Discourse on a New Method: Reinvigorating the Marriage of History and Philosophy of Science

Discourse on a New Method Reinvigorating the Marriage of History and Philosophy of Science EDITED BY

Mary Domsl(i and Michael Dicl(son WITH A CONCLUDING ESSAY BY

Michael Friedman

OPEN COURT Chicago and La Salle, Illinois

[lJ Introduction Discourse on a New Method) or a Manifesto for a Synthetic Approach to History and Philosophy of Science MARY DOMSIZI and MICHAEL DICIZSON

Questions about the marital status of history of science and philosophy of science are older than the flurry of new departments, centers, and programs in the history and philosophy of science that emerged in the midtwentieth century. 1 But quite naturally their creation, and the ongoing joint meetings of the History of Science Society (HSS) and Philosophy of Science Association (PSA), have propelled these questions to the foreground on several occasions. One such event took place at the University of Minnesota in 1969, at a conference dedicated to the rationale of the union of history of science and philosophy of science. 2 The conference papers were published under the title Historical and Philosophical Perspectives of Science (1970), as Volume 5 of the Minnesota Studies in the Philosophy of Science (edited by Roger H. Stuewer). A few years later, Ron Giere wrote a provocative review of the book, entitled "History and Philosophy of Science: Intimate Relationship or Marriage of Convenience?" (Giere 1973).3 Giere considers various arguments for the necessary union of these disciplines-arguments put forward by an esteen1ed group that includes Herbert Feigl, Paul Feyerabend, Mary Hesse, Ernan McMullin, and Arnold Thackray-and, on the whole, he finds their appeal to history as a remedy for the failings of the logical empiricist movement in the philosophy of science unconvincing. Furthermore, Giere does not simply find fault with the arguments presented in the Minnesota Studies volume; he suggests that, although the pairing of history and philosophy of science had already become fashionable, contemporary philosophers of science needn't worry themselves with appeal to the historian's 'internalist' studies of scientific development or scientific practice. Attention to what he terms "real science"-the science that can be learned in textbooks-will serve their philosophical purposes. As such, he argues that the union of these disciplines "lacks a strong 1

2

Mary Domski and Michael Dickson

conceptual rationale," and thus the n10st (or best) we can say is that the history of science and the philosophy of science share a common interest in science (Giere 1973, 296). Unsurprisingly, Giere's forceful remarks have been met with equally forceful replies from historically-minded philosophers. Ernan McMullin offered his direct response to Giere in a 1974 PSA paper entitled "History and Philosophy of Science: A Marriage of Convenience?,,4 McMullin argued that, despite Giere's fervent suggestions to the contrary, there are in fact particular issues in the philosophy of science that cannot be adequately addressed without remaining sensitive to the history of science, and especially to the historical development of scientific practices and scientific theories. The principal issues he cites concern the assessment of theories, the nature of scientific growth, and the ontology of theoretical entities. He aims to show that if "an adequate treatment even of the n10st 'logical' of these three questions, viz., that of theory-assessment, cannot avoid reference to the history of science," then we may safely conclude that the "n1arriage of the history and philosophy of science is not just one of convenience" (McMullin 1974, 586). Even without a thorough assessment of McMullin's argument, our current state of practice in the philosophy of science indicates that McMullin and company have won the debate. For there is no question that philosophers of science have in the past few decades made it commonplace to bring the history of science and the history of the philosophy of science to bear on their philosophical discussions, and they have done so quite effectively. For instance, historical examples have figured prominently in some of the more compelling arguments concerning the nature of scientific growth. Larry Laudan's (1981) anti-realist appeal to the 'pessin1istic n1etainduction' draws its force from revolutionary periods in the history of science' and John Worrall's (1989) 'structural realist' reply earns credibility from what he understands as Poincare's understanding of how scientific theories represent the world. 5 So much the worse for Giere's arguments it seems, and so much the better for historically-minded philosophers (and philosophically-minded historians) who can appeal to the continued and successful application of the history of science in the philosophy science as justification that these disciplines are intimately related. However, the debate cannot be settled on these grounds alone. For when Giere claimed that the history of science has no significant bearing on the philosophy of science, he was not claiming that philosophers should ignore the science and scientific practices of the past. The argument was hardly this simple-minded. Giere was instead, and more pointedly, arguing that the sort of history that historians ofscience tal(e as paradigmatic to their discipline-a fine-grained and often complex history, focused, more or less, on the specific events, context, and influences surrounding the develop-

Introduction

3

ment and acceptance or rejection of a concept or theory-is not the sort of history from which philosophers can prima facie draw usefullessons. 6 Much philosophical use of history (e.g., as exercised by Laudan) does not involve history of that sort. Surely Giere would have had exactly this point made to him by Edward Grant, Richard ('Sam') Westfall, and his other historian colleagues at Indiana University at the time. For Giere, then, the fact that our contemporary philosophical discourse is peppered with appeals to and examples from the past hardly counts as genuine evidence that the history of science and the philosophy of science are involved in an intimate relationship. Any such evidence for this relationship will come from a more reflective consideration of the conceptual ties between these disciplines. In particular, Giere urges that historicallyminded philosophers "not be content merely to practice their art but . . . make repeated efforts to explain and argue the rationale for their approach" (Giere 1973, 291), and he solicits from historian-philosophers an account of how "philosophical conclusions may be supported by historical facts." "Until this is done," he writes, "the historical approach to philosophy of science is without a conceptually coherent program" (1973, 292). Our aim in this 'manifesto' is in part to provide the sort of historiographical and methodological account that we think Giere rightly urges. The account we offer talces its inspiration from the historiography that has been masterfully applied by Michael Friedman over the past few decadesa historiography which we have dubbed a "new method" in the title of this book. Though Friedman is not the only historian-philosopher to apply the methodology we describe and encourage, we believe that attention to his course of work grants us important insights into how best to understand the complexity and intimacy of the relationship between the history of science and the history of philosophy. Our intention in this volume is to explicate and pursue these insights (though certainly not uncritically), and thereby provide a means by which to reinvigorate the very marriage of the history and philosophy of science that Giere and others have continued to bring into dispute.

"Remarl(s on the History of Science and the History of Philosophy"7 Some twenty years after Giere's review was published, Friedman (1993) made what perhaps seemed at the time a modest plea to contemporary philosophers of science. Taking for granted that Thomas IZuhn's Structure of Scientific Revolutions (1962) had convincingly shown that philosophers of science ought to take seriously the historical development of science, Friedman suggests that historians of philosophy ought to take seriously the

4

Mary Domski and Michael Dickson

historical role that science has played in philosophy. They should, that is, consider the historical interaction between the history of science and the history of philosophy. Relying on IZuhn's historiography of science, and adopting a more or less Kuhnian characterization of scientific revolutions, Friedn1an argues, in particular, that an important task for the historian of philosophy is to situate the emergence and evolution of philosophical ideas in their revolutionary scientific context. Friedman does not have in mind what he rightly supposed to be, by then, a commonplace in the history of philosophy, namely, the recognition that broad themes from the history of science-and especially the largescale transformations in philosophical concerns that often accompany scientific revolutions-have some part to play in our understanding of the history of philosophy. In this vein, Friedman mentions the philosophical concerns of figures such as Descartes, whose work can be understood in the context of the shift from Aristotelian-Scholastic natural philosophy to the mechanical philosophy. Other examples spring to mind: the role of the systematization of geometry in the development of various ancient Greek accounts of knowledge; the role of the developn1ent of computer science in the development of various theories of mind; and the role of the shift from classical to quantum theory (together with the perceived 'acausality' and attendant 'irrationality' of the latter) in the development of various twentieth century 'irrationalist' philosophies. Recognition of such broad parallels between the history of science and the history of philosophy (and corresponding discussion of the direction of causality, if any) is widespread. Without discounting the importance of such broad-brushed considerations, Friedman has in mind a finer-grained discussion of the connections between science and philosophy, one that takes account of the specific content of, and especially the central problems driving, the science of the day. More specifically, he emphasizes that many philosophical enterprises are motivated by the attempt to understand (and in some cases provide) the detailed foundations of scientific theories. In his 1993 paper, Friedman explores the possible fruits of adopting this historiographical perspective in the context of two examples. The first concerns the change from an Aristotelian-Scholastic physical dynamics to a mechanical, ultimately Newtonian, dynan1ics-a change motivated by the work of figures such as Galileo, Descartes, and, of course, Newton himself. That shift, argues Friedman, raised new foundational problems in science that in turn gave rise to philosophical questions concerning the status and possibility of our knowledge of the natural world. Crucial among the problems emerging from Newton's system of the world was our ability to distinguish states of true motion from states of true rest and, perhaps even more important, states of inertial motion from states of non-inertial (accelerated) motion. Specifically, it is unclear how the new dynamics of the

Introduction

5

Principia Mathematica has any empirical content, in part because the theory is Galilean-invariant, so that there is nothing given (such as an Earth at absolute rest, as there was in the older system) relative to which, for example, inertiallTIotion is defined. 8 But then how does Newton's law of inertia (a law about inertial bodies) have any empirical content? (The second law does not help here, unless one already knows all of the forces affecting the systelTI, a point that would much later push Poincare towards his conventionalism about geometry.) Friedman suggests that the dialogue that ensued between science and philosophy in the wake of the Principia's success and acceptance centered on "this fundamental problem lying at the basis of the new physical dynamics" (Friedman 1993,41). And this suggestion motivates reconsideration of how to best understand the course of modern thought. For if we take seriously the interaction between science and philosophy and, in turn, approach the evolution of the modern philosophical tradition as successive attempts to come to terms with the fundamental problem of relativity of motion that lay at the basis of the new physical dynamics, then, on Friedman's account, we are brought to a conception of the evolution of modern philosophy that is more illuminating than the conventional picture of a succession of largely futile attempts to solve the problems of skepticism about the external world. (43) This re-evaluation of n10dern thought, centered on the revolutionary character of the history of science, sheds new light on the progress of epistemology and metaphysics. Specifically, Friedman claims that contrary to the opinion of some contemporary historical writers, the philosophers of the modern tradition are not best understood as attempting to stand outside the new science so as to show, from some mysterious point outside of science itself, that our scientific knowledge somehow 'mirrors' an independently existing reality. (48) Rather, as Friedman argues, they are best understood as trying to come to terms with the foundations of the new science. IZant, for Friedman, is paradigmatic here, because he proposes that metaphysics should articulate the a priori conditions of the possibility of experience. Starting from the so-called synthetic a priori forms of knowledge expressed in Euclidean geoll1etry and Newtonian dynamics, lZant sought not to justify this knowledge but to investigate those very conditions of human cognition that make such synthetic a priori knowledge possible. By grounding the project and goal of metaphysics in mathematics

6

Mary Domski and Michael Dickson

and natural science, IZant at the same time established "a reorganization and reinterpretation of metaphysics itself" (49), and thus instigated his own 'revolution', not in science, but in philosophy. On Friedman's account, then, Kant's great service to philosophy was to bring metaphysics down from its lofty place 'above' human knowledge precisely by taking pure mathematics and pure natural science as the starting point of his metaphysical investigations. In his second example, Friedman points out that a similar concern with our most successful claims to scientific and mathematical knowledge emerges in the early twentieth century as philosophers faced another revolution in the natural sciences: the special and general theories of relativity grounded on non-Euclidean geometry. Remaining sensitive to their historical situation-and, in particular, to the ongoing scientific revolutions of the early twentieth century-Friedman contends that we can understand logical positivists as attempting philosophically to come to terms with the profound conceptual revolutions that initiated twentieth-century science. These thinkers should be seen not as attempting to justify twentieth-century science from son1e sterile and futile external vantage point but rather as once again refashioning the basic concepts and principles of philosophy so as to accommodate and comprehend the new scientific developments. That is, their aim is not to justify twentieth-century science from some supposed "higher" standpoint but rather to provide a rational reconstruction of that science and to find thereby a new, non-metaphysical task for philosophy. (49) With the supplanting of Newtonian dynamics and its framework of Euclidean geometry-and thus with the possibility of scientific knowledge as understood by IZant no longer on center stage-the logical positivists sought to trace the development of scientific ideas and scientific theories by embedding the development of scientific theories in a logical language in order to exploit the rational interconnections among our scientific claims. However, although the protagonists had, out of historical necessity, abandoned certain aspects of IZantianism in its original form, the spirit of IZant's project for metaphysics was still alive during the early twentieth century. In particular, Friedman argues that the logical positivists continued to "advocate a modified lZantian position according to which there is a fundamental distinction between the spatiotemporal framework of physical dynamics and the empirical laws formulated within this framework" (50). This fundamental distinction is not simply a IZantian distinction between the a priori and the a posteriori; faced with revolutionary changes in math-

Introduction

7

ematics and science, it was transformed into a distinction between the relativized a priori and the a posteriori, between historically dynamic matheInatical and physical laws on the one hand, and the understanding of the world that depends constitutively on then1 on the other hand. Recognizing this IZantian thread running through early twentieth century philosophy of science, Friedman concludes that although the logical positivists' preoccupation with the a priori did indeed thereby preclude them from using the history of science as a philosophical tool, this did not prevent them from recognizing the profound philosophical significance of conceptual revolutions in science. On the contrary, their effort to articulate a coherent conception of the relativized a priori must, I think, count as the most rigorous attempt we have yet seen philosophically to come to terms with precisely such conceptual revolutions. Of course, as we have also seen, this heroic attempt of the logical positivists was in the end a failure. Yet I do not myself think that we will ever progress beyond this point until we possess a fuller appreciation of the historical evolution of our own philosophical predicament. And this means, as I have tried to emphasize throughout, that we must attend more closely to the history of science' the history of philosophy, and to the essential interaction between them. (54) In these concluding remarks, Friedman proposes a project not only for historians of philosophy, but also for philosophers in general. Note, in particular, the word 'essential' in the final sentence. Friedman's implicit claim is that the historical interaction between philosophy and science is not a quirk of history, but an essential fact about the two disciplines. It is of the nature of science and philosophy to interact. Without placing too much weight on this point, or seeking to follow through on this idea in detail (which in any case would require a long discussion of the nature of science and philosophy), we must nonetheless emphasize the distinction between the claim that philosophy has, as a matter of historical fact, interacted with science in a variety of ways (as exemplified in Friedman's two examples), and the claim that philosophy by nature does, and ought, to proceed in that manner. Friedman quite clearly endorses the former claim, and appears to endorse the latter as well. Indeed, the two claims are not entirely independent. Intellectual disciplines do not exist in an historical vacuum. The standards, problems, goals, and methods of a discipline provide the context that malces its practices intelligible, and they are in part inherited from its own past. In the case of philosophy, insofar as this context arises from substantial historical interaction with science, understanding that historical interaction is required for

8

Mary Domski and Michael Dickson

understanding our present practices. Hence, while the two claims that we have thus far attributed to Friedman are indeed distinct, the first does give reason to believe the second. Nonetheless, one might say that Friedman has taken the role of science in the history of philosophy to an unhealthy extreme. The argument runs thus: Remaining so firmly wedded to a scientific perspective on, say, the modern period provides us at best only half the picture. For example, skepticism was in fact in bloom and helped shape epistemology from Descartes to IZant and beyond. Though initially persuasive, notice that this sort of argument is grounded on a worrisome either-or: either we examine how the history of science has influenced the history of philosophy, or we examine how skepticism influenced the history of philosophy. We certainly do not deny that Friedn1an has adopted the history and philosophy of science as the focal point of his investigations. But, as his recent work has illustrated, adopting this strategy does not force us to thereby ignore other factors that contributed to the development of novel philosophical insights. For Friedman's project is not simply one of illuminating how the history of philosophy has helped shape the history of the philosophy of science, and vice versa. Rather, by turning our focus toward the history and philosophy of science, he has shed new light on the standard intellectual concerns surrounding skepticism, metaphysics, and theology that shaped (and continue to shape) the historical progress of philosophy.9 Thus, what some might take as the overly radical nature of Friedman's work we talce, more sympathetically, as a strategy for drawing attention to an interaction between philosophy and science that has, until very recently, attracted inadequate scholarly attention, as well as a fruitful strategy for giving old questions new flavor.

A Tradition of Historical Philosophy and Philosophical History There is more to Friedman's story than indicated from the above sketch of his 1993 article. From a plea to talce seriously the interaction between the history of science and the history of philosophy, Friedman arrives not only (apparently) at the conclusion that philosophy and science ought to interact in a certain way, but also (clearly) at a IZantian, or better, neo-IZantian, conception of the nature of that interaction, which rests on the relativized a priori. IO If one finds the science-influenced story that Friedman tells hard to digest, then his IZant-influenced story about the history of science and the history of philosophy may be poison for the palate. But we need to take care, as always, not to throw the baby out with the (IZantian) bath-

Introduction

9

water, and ask whether embracing the sort of historiography of philosophy that Friedman promotes in fact demands that we at the same time embrace a IZantian or neo- IZantian understanding of the growth of scientific knowledge and the methodology of philosophy. From our perspective, the two are not inextricably bound. In fact, there is a long tradition of pursuing the sort of historically informed philosophy and philosophically informed history that has become a hallmark of Friedman's work. Seeing Friedman's own work in this context will help us to understand, first, how specifically he differs from his predecessors, and also how one might pursue a program broadly sympathetic to Friedman's approach without necessarily ending up a neo-IZantian.l 1 On the side of history, Friedman stands in a tradition of philosophical historians that includes Richard Westfall, who approached the historical development of ideas with an eye to the philosophical issues that loomed over seventeenth-century science. Westfall's Force in NewtonJs Physics (1971) provides an example. In his masterful explication of Newton's formulation of the second law of motion, Westfall puts less emphasis on the attempt to formalize the law (F= rna) and more emphasis on the philosophical context that allowed Newton to ask the right questions in the first place. It was Newton's proposed resolution to "the metaphysical question of the ontological status of force" that, according to Westfall, brought him to the notion of force that lay at the heart of his mature physical dynamics. In a similar fashion, Friedman has brought us to a deeper appreciation for the philosophical questions that enabled Einstein to develop relativity theory. On Friedman's account, Einstein's attention to the philosophical questions surrounding the newly discovered non-Euclidean geometries, as illustrated by his close reading of Helmholtz and Poincare, partly allowed him to develop a theory of relativity that stands on the foundation of a non-Euclidean geometry (cf. Friedman 1999, 2001). On the side of philosophy, Friedman attests to his own position in a tradition of historical philosophers, most notably, Thomas IZuhn. Quite clearly, their respective philosophies of science take revolutionary changes in the history of science as their touchstone. Where IZuhn (and followers) see, in a scientific revolution, an opportunity to draw in1portant generic philosophical lessons about general philosophical issues, such as the nature of scientific knowledge, Friedman sees in them an opportunity to understand the specific problems that drove the development of philosophy at the time of the revolution and beyond. Friedman also draws significant motivation from another historicallyminded philosopher, Ernst Cassirer. Cassirer famously attempted to preserve a lZantian understanding of science and knowledge in the wake of the discovery of non-Euclidean geometry and Einstein's theory of relativity. Relying heavily on IZant's notion of "the regulative use of reason," Cassirer

10

Mary Domski and Michael Dickson

presented a (Marburg neo- I No

S are P.

Aristotle did not fail to see the formal validity of such an argument, but such a syllogism made no sense in terms of his conception of reality. Given that the category of substance guided Aristotle's thought, this concept could not be ignored even in logic. 21 To recognize a fourth figure syllogism like the above demands that, in the premises, we regard a substance (the major term, S) to be the predicate of some characteristic (the minor term, P). That would be an absurdity in Aristotle's metaphysics, yet the fourth figure exhibits such an order. Hence, from the standpoint of Aristotle's metaphysics, there was good reason not to include the fourth figure in logic. While Kant could still claim that logic was complete-a sphere in which there were neither steps ahead nor backwards since the time of Aristotle-the sphere of logic in fact underwent a transformation when Boole, Peano, Peirce, Russell, and Whitehead reinterpreted logic in genuinely formal terms as a system of possible relations. Valid inference was no longer interpreted with any reference to existing things. This transformation of logic stands behind Cassirer's first revisions of Kantianism (Cassirer [1907] 2001 and [1911] 2001). The problem of the validity of non-mathematical cultural forms is not how to unifY or even to integrate them with the kind of validity found in mathematics, but rather to understand how such different kinds of meaning can be related at all-how the expressive, representational, and purely significative functions hang together. This, I take it, is what stands behind Friedman's claim that Cassirer never satisfactorily explains his outlook.

272

John Michael J(rois

6. Concluding Observations Friedman's criticism is more fundan1ental than he intends it to be. For he points to the fact that Cassirer left his most fundamental conceptions inadequately expressed. The most important topics in Cassirer's philosophy of symbolic forms are symbolism and form, but one must look long and hard for his explications of these notions (Krois 2002a). His views can be reconstructed, but they are not obvious. In Cassirer's recently published lecture on "Probleme der IZulturphilosophie," he indicates explicitly that the cultural sciences are based upon the concept of "form," not "value" as Rickert claimed (Cassirer 2004, 101). He goes further and claims that philosophers have failed so far to come to grips with cultural concepts because they lacked a proper conception of their "logical place" (logischer Ort). This, he says, is the concept of style: "All cultural concepts are concepts of style" (Cassirer 2004, 103). Instead of seeing particular phenomena in reference to a causal law, they are exemplary of a style. Hence, a particular object-an unknown drawing-is attributed to a time (baroque), a place (Netherlands), an artist (Rembrandt), and a period (late Rembrandt) by a series of style concepts. The point of the cultural sciences is not to subsume particular phenomena under universal value headings such as the good or the beautiful, but to make sense of these particulars, either in regard to their place in history as representatives of particular "forces" or changing styles. 22 The meaning of an individual action (Cato's suicide) or of an object (Raffael's school of Athens) is determined by the extent to which it is possible to relate it to different narratives, or to concepts of style (Cassirer 2002, 14-17). These permit us to "characterize" it as somehow "exemplary" of a style, a political conception, or a worldview (Cassirer 2004, 63f.). This characterization does not aim to derive the phenomenon in question from any law (Cassirer 2004,73), but to elucidate it in terms of its organization, which can be differentiated from comparable styles or worldviews. Hence, historical concepts such as "medieval" or "renaissance" make use of idealized characterizations that allow us to describe a type of historical personality (the typical renaissance man), but the individual figures that illustrate such a characterization (Leonardo, Machiavelli, Cesare Borgia) cannot be subsumed under any kind of law (Cassirer [1942] 2000,70-73). Cassirer says that such notions can be "universally communicable" without being necessary (Cassirer [1942] 2000, 118). Objectively valid knowledge in the cultural sciences depends upon the objective validity of the phenon1ena studied by these sciences: the styles or formulae used. The objectivity of mathematical relations is clear, but the objectivity of specific cultural concepts is always open to debate. These difficulties appear to increase insofar as cultural phenomena involve the trans-

Validity in the Cultural Sciences?

273

ll1ission of feelings, which seem to resist even this kind of formal understanding. Yet Cassirer claimed that even expressive phenomena possess objective validity. Cassirer's success or failure in treating the cultural sciences ultimately depends upon his treatment of the objectivity of the expressive function of meaning. This brings us back to symbolic pregnance, which is supposed to mediate between every type of meaning: expressive, representational, and purely significative. Phenomena possess different kinds of symbolic value: qualitative (say, a perceived blue), representational (the color type "blue" as named by the English word), or significative-a scientific object not directly visible to the naked eye (a particular wavelength in the spectrum of electromagnetic waves). The problem is how the first kind of, seemingly private, phenomenon can be objectively valid. Cassirer cautioned that the usual examples of perception in philosophy, such as a "red patch" or "blue spot," are not representative of actual perception. In reality, we normally perceive colors expressively, say as a "cold blue" or a "hot red." Cassirer pointed to empirical research that showed that expressive phenomena are more primitive than the philosophical concept of a simple perception (such as a blue spot) (Cassirer [1929] 2002,76, [trans. 65]). Phenomena such as perceiving a "cold blue" color are actually so commonplace that they go unheeded by philosophers. According to Cassirer, this type of multisensory perception is actually the rule, not the exception-a claim that has been confirmed in the most recent research. 23 Cassirer claimed that the spheres of visual and auditory sensations, of smell and taste, are more closely linked together than we usually recognize because of our tendency to isolate specific 'qualities' in thought (Cassirer [1929] 2002, 41 [trans. 34f. J). Cassirer asserted that the separation of the senses is a derived, not a primitive phenomenon. We only arrive at the data of distinct discrete sensation-such as light or dark, warm or cold, rough or sn100th-by setting aside a fundamental and primary stratum of multisensory perception and eliminating it, for theoretical purposes (Cassirer [1929] 2002, 85 [trans. 73J). This stratum-the perception of expression (Ausdruckswa,hrnehmung)-is composed of phenomena such as what we describe in phrases like a "cold blue," a "sharp odor," or a "deep tone." This kind ofperception, Cassirer stresses, is "at first a mere passivity, a being-acted-upon rather than an acting" (Cassirer [1929J 2002, 88 [trans. 75J). The phenomena themselves appear agitating or soothing, gloon1Y or joyful, pacifying or terrifying. This perception of expression can be diffuse, such as the perception of a mood or emotional atmosphere, or it can be focused, such as in the recognition of a physiognomic character-a sn1iling or frowning face, but it is above all an objective phenomenon. Naturally, an individual with a "lively imagination" is able to see what others do not, recognizing faces in clouds or hearing voices in the moaning of the wind. But others

274

John MichaelI(rois

could perceive these phenomena as well. Cassirer's point was that the perception of expression as such was a type of symbolic phenomenon with a potential generally validity. Cassirer did not complete his philosophy of the cultural sciences nor have today's cultural theorists managed to develop a science of culture recognizing the objectivity of expressive syn1bolism. The recent reintroduction of the study of feeling in neuroscience and the second generation of cognitive science with its emphasis on embodied intelligence, are perhaps indications that the time has come for an interdisciplinary research program like Cassirer's which sought to link natural and cultural sciences. The general recognition in cultural studies of the fundamental importance of symbolic processes is a step in this direction. Cassirer believed that objective validity in the cultural sciences hinged upon understanding the symbolic character of expression. Many of his writings on the subject have only recently appeared and some the most important, especially his texts on "symbolic pregnance," are still forthcoming (Cassirer in preparation). It is still too early for a final judgment on his success or failure in showing the comparability of the natural and cultural sciences. But I hope to have shown the direction in which to look for a solution to the question: in the concept of symbolic pregnance.

NOTES 1. I want to thank the editors of this volume, Mary Domski and Michael Dickson, for providing me with many helpful criticisms of the first version of this paper. 2. "Werte sind keine Wirklichkeiten, weder physische noch psychische. Ihr Wesen besteht in ihrer Geltung, nicht in ihrer Tatsachlichkeit." 3. "Die Natur erldaren wir, das Seelenleben verstehen wir." 4. "Die Gesamtheit der Objekte, an denen allgemein anerkannte Werte haften, und die mit Riicksicht auf diese Werte gepflegt werden." Cf. also Rickert (1910), 89. 5. For Cassirer's criticism of Rickert, see e.g., Cassirer ([1942] 2000), 37; for his criticisn1 ofHeidegger, see Cassirer (1995), 219-24 [trans. 200-208]. 6. Friedman (2000), 34-37 gives a further account of Cassirer's criticisms of Rickert. 7. Hamlin and I
423

424

Scott Tanana

geometry, a dynamics presupposes a kinematics, and so on. A theory simply is not well defined without a specification of a particular framework, and so the principles of the framework are a priori in the second sense. But while some set of a priori principles is required to provide a framework in which the theories can be specified, no particular, given set is required; we have a choice regarding the form of that framework. Similar phenomena can be described by theories with different frameworks. For example, consider a phenomenon in which all objects in the absence of specific forces move along certain curved lines. Such a phenomenon might be described by a theory with a flat geometry and universal forces, or by a theory with a curved geometry, where what would have been forced motion in the first framework is now considered to be free motion. For Friedman, the choice to adopt a new theoretical framework may be based on a variety of factors, including well-articulated meta-principles and desiderata for a theory as well as empirical considerations. However, I believe that Friedman has a difficult time accounting for the role that empirical phenomena play in such a choice, because while the phenomena are supposed to be concrete and independent of the abstract framework, they must also be general and well defined, which is possible only in the context of such a framework. According to Friedman, the abstract mathematical formalisms of modern physical theories have no empirical content on their own. An abstract mathematical apparatus only makes claims about empirical phenomena once one indicates how the abstract structure is to be applied to empirical phenomena. One indicates this connection via what Friedman calls (following Reichenbach) "coordinating principles." These principles coordinate "mathematical representations with experience." That is, they connect elements of the abstract structure to empirical phenomena, and thereby give the abstract theory empirical content: [Coordinating principles] serve as general rules for setting up a coordination or correspondence between the abstract mathematical representations lying at the basis of [the physics] . . . and concrete empirical phenomena to which these representations are intended to apply .... And without such general rules of coordination we simply have no idea what it means for concrete empirical phenomena to be described by the mathematical representations in question-either correctly or incorrectly. (Friedman 2001, 76-77) Friedman's paradigm of a coordinating principle is the light principle in special relativity, the principle that the speed of light is constant in any inertial reference frame. 2 Minkowski space-time by itself is merely an abstract mathematical forn1alism. The light principl~~~~~_cj~~~~_~~_~952~4ig~!~~_what----

Theory) Coordination) and Empirical Meaning in Modern Physics 425 are otherwise just abstract elements of Minkowski space-time with concrete empirical phenomena. In particular, the light principle coordinates the null straight lines of the former with possible paths of light rays in the latter, and the angle of those lines in a chosen space-time basis with the speed of light in a reference frame. When null straight lines are taken to correspond to possible paths of light, and coordinatizations 3 of the Minkowski formalism are taken to correspond to the coordinate systems of inertial reference frames, then the conformal structure of Minkowski space-time indicates that the speed of light will be constant and independent of inertial reference frame. That coordination then tells us in general how to apply the Minkowski structure to the empirical phenomena of observed distances and times in reference frames, or to translate the observed phenomena in a reference frame into Minkowski space-time terms. Until we do this, Minkowski space-time is not even a space-time and does not indicate anything about any phenomena, never mind specific phenomena concerning light, length contraction, or time dilation. 4 In the old Kantian picture, such coordination is not needed, as our experiences immediately and irrevocably occur in a particular a priori framework. In particular, Euclidean geometry is the only possible spatial form of our experience for Kant, and our experience of the world cannot at all be isolated from that form. Friedman rejects the necessity of Kant's particular Newtonian-Euclidean framework, but like the early logical positivists, he maintains that some choice is required, and he retains the idea that the a priori, once chosen, is constitutive of the objects described by theories (Friedman 2001, 2002). That is, the characteristics of the theory depend on the principles themselves, and so a conceptual framework establishes the form and nature of those objects in the framework. 5 Because of this constitutiveness, the objects of the framework, as objects ofthe framework, cannot be isolated from the form of the framework-theoretical objects are deeply embedded within a particular framework and inherit the structure of the framework of their embedding. That is, the things described by a theory, as theoretical objects conceived of and described by the theory, have structures and properties that depend on the framework involved. This constitutiveness thus rules out a naive reading of what it means to be "described by a theory" that would allow that one and the same object can be described differently by different theories. The objects and events of a classical space-time are different from those of a Minkowski space-time because the frameworks are different. 6 However, this constitutiveness cannot be all encompassing, for on Friedman's view we can choose among different coordination procedures that would coordinate the empirical phenomena of experiences to different frameworks. If we have a choice of the form of the scientific characterization of our experiences, then apparently there is something about

426

Scott Tanona

those experiences that is independent of the form prior to our choice. So apparently we, unlike IZant, must reject the aspect of the a priori that makes the form and content of our experiences inseparable. 7 Coordinating principles (at least implicitly) must refer both to an abstract structure and to the "concrete empirical phenomena" of our experiences, as they provide the abstract structure empirical content by indicating how they are related. But if this is so, the empirical phenomena must be specified independently of that framework. If the phenomena were already constituted by the framework, the coordinating principles would not be necessary. If they were not able to be specified independent of the framework, the principles would be empty. The separability of form and empirical content suggests a difficulty not well handled on Friedman's picture. In the early discussions between Schlick, Reichenbach, and Cassirer on the idea of coordination (Zuordnung), there was much disagreement about what it is to which a priori structures are coordinated (see, e.g., Ryckman 2005, ch. 2 or Coffa 1991, ch. 10). In general, the idea was that the conceptual is coordinated with the physical world. But what is this world? Is it a pre-existing reality which becomes designated by our signs, some set of elements from our experience which becomes conceptualized through the framework, or some other sort of "undefined" which is only defined through coordination? Friedman's interest, like Reichenbach's, is not in the individual nature of cognition but rather in more general features of scientific epistemology. Friedman's idea may seem to be that the concrete empirical phenomena are something like our direct observations or perceptual experiences. For example, in one place he describes the coordinated as "concrete sensible experience" (Friedman 2001, 76). However, his actual examples are much more broad and intersubjective than direct personal sense experiences. One example, of course, is the phenomenon of the constancy of the speed of light; another concerns "the observable relative motions in the solar system" (Friedman 2001, 76-77). Both are indicative of Friedman's broader claim that the coordination of the relativized a priori concerns the non-personal, intersubjective facts of "concrete empirical phenomena." Like Reichenbach, Friedman therefore needs to account for the independence of the empirical phenomena in non-personal, intersubjective terms. However, this is a difficult task on Friedman's conception of the relativized a priori alone, and one I believe he has not yet addressed. I suggest that modifying his conception will help in this task. I will argue below that abstract theories gain empirical content through coordination to phenomena defined in other frameworks. To develop this view I will focus on special relativity, the empirical phenomena of which, I will argue, are based in a classical framework. I will then apply this perspective to quantum mechanics and suggest that the use of classical concepts in quantum

Theol'Y, Coordination, and Empirical Meaning in Modern Physics 427 mechanics is required and natural, even though the justification for their use is not as well developed as it is for relativity.

2. Theory-Ladenness and Other-Theory-Ladenness What, then, are the empirical phenomena to which abstract theories are coordinated? In particular, what are the phenomena of special relativity? One sometimes hears that the empirical or observational content of relativity is specified via definitions of space and time in terms of operations with rods and clocks. But operationalism is far too limited a picture to account for the empirical content of theoretical terms. Our concepts of distance and time play many roles in a wide range of experimental and theoretical settings. Corresponding to these many roles are many ways of measuring distance, time, and speed. For instance, we can measure the speed oflight by using rods and clocks, or (as we shall see below) through interference effects, or through any variety of other means. We can measure distance in a variety of ways that do not involve rods. And we can measure time with pendula, digital clocks, cesium atoms, quasars, or a variety of regular periodic phenomena. The empirical claims of relativity are independent of one's method of measurement: time dilation is not merely a slowing of clocks, but is a slowing of all time-dependent phenomena. The problem is that there does not appear to be a good operational justification for why we take all of these different ways of measuring distance and time to be measuring the same things (respectively), as it is not clear that they are at all operationally intertranslatable. But it is clear that many different operations are understood as measuring the same thing. Duhem expresses this idea very well: Suppose the following sentence is pronounced to a physicist: "if we increase the pressure by so many atmospheres, we increase the electromotive force of a battery by so many volts." It is indeed true that the initiated person who knows the theories of physics can translate this statement into facts and can do the experiment whose result is thus expressed, but the noteworthy point is that he can do it in an infinity of different ways. (Duhem [1906] 1991, 148-49) Duhem continues to say that one could, for instance, measure the change in the electromagnetic force through any of a variety of galvanometers, voltmeters, and so on. It is our scientific theories that tell us that these different operations measure the same things. Neither operationalism nor any pure observation language can provide the empirical content of a theory, as the empirical

428

Scott Tanona

content of a theory is not restricted either to particular operations or to particular, direct, perceptual observations. Part of the reason for the fall of logical positivism was the growing consensus that observations in science are theory-laden and that there is no clear observational/theoretical distinction. So while there are many facts about the operations one can perform on light, and there are many different observations one could make in the process of measuring the speed of light, there is but one fact about the speed of light, and it is the latter fact that embodies the empirical content of special relativity. Now, theory-Iadenness often is thought of as a claim about the psychology of perception. As such, it is the claim that when we observe some phenomenon, what we actually see and how we literally perceive it depends on background theories and beliefs. One special class of cases are optical illusions, where two people (or the same person at different times) may receive exactly the same sensory input from a purposely ambiguous line drawing but see different things (e.g., an old woman and a young woman). The classical example of this sort of theory-ladenness of phenomena from IZuhn is the deck of cards that has a red six of spades and a black four of hearts (IZuhn [1962] 1996, 62). When shown these cards quickly in succession with other cards from the deck, people tend not to notice anything wrong because, the argument goes, we unconsciously and immediately interpret ambiguous sensory input in accordance with expectations determined by background "theories" about decks of cards. These examples and others like them are meant to show that our perceptual processes can be affected by many things, including our expectations about a particular setting. IZuhn went so far to say in Structure that a scientist used to seeing cloud chamber tracks "sees" a particle where the untrained eye sees only tracks ([1962] 1996, 195). However, there is a broader but related notion of theory-ladenness that concerns a broader notion of "observation" in a loose sense. 8 This notion may also be present in IZuhn's Structure, but it is not related to the actual psychological process of perception. It rather concerns the empirical aspect of theory investigation in the sense in which I interpreted Duhem's quotation above. In this sense, a scientist may make many "observations" in a lab that are only indirectly related to her perceptual systems and which involve much explicit interpretation of the workings of her instruments, inferences from other "background" theories, and so on. This sense of observation is meant to be loose specifically because it suggests that the en1pirical investigation of a phenomenon, even if for the explicit purpose of testing a theory, itself involves the conscious and deliberate use of much theory, so that the analogue of observation in this conception of empirical investigation is fairly disconnected from the actual specifics of any perceptual process.

Theory) Coordination) and Empirical Meaning in Modern Physics 429 For example, if a theory predicts that the current in a wire will have a particular value under certain circumstances, we can test the theory by "observing" the current. Of course, a scientist does not directly see the current, and she need not even see the ammeter being used to read off the current to get the result (perhaps she is read the values by a computer-generated voice). Independently of how she learns the value of the current, her "observations" of the current are being used to develop evidence for the theory. There are reasons for preferring this notion of "observation" in science, precisely because observation in the scientific setting ought to be intersubjective-something all potential participants would agree to-and direct perceptual sensation is just not amenable to that. Thus the term "observation" loosely refers to the recording of empirical facts thought to be relevant for a particular theory. The idea of theory-ladenness of observations in this sense of "observation" is the idea that the description and even conceptualization of the intersubjective results of an empirical investigation involve a range of background theories which we often talce to be unproblematic for the purposes of the experiment or measurement at hand but which nevertheless are not neutral. Theory-ladenness may pose a problem if the theory being invoked in the measurement of a phenomenon is the same one being investigated as an explanation of the phenomenon. Indeed, this is the alleged issue in paradigm non-commensurability. However, this problem is avoided if the theories involved in "observation" of phenomena need not be the theories being used to explain or predict those phenomena. Circularity problems can be avoided if observations are other-theory-laden. How are these issues relevant to Friedman's conception? This second sense of theory-ladenness is analogous to the idea that a conceptual framework is required for the conceptualization and description of empirical phenomena. In a footnote discussing the problems with operational definitions, Friedman indicates how coordinating principles, instead of operational definitions, can be used to define the basic concepts concerning measurements of empirical phenomena: [I]f we define the uniform passage of time by stipulating that some actual periodic process (such as the diurnal rotation of the earth) is uniform, this would be an operational definition of "equal times." Defining "equal times" by the laws of motion, by contrast, explicitly provides for the possibility of correcting and refining any and all such concrete coordinations without limit. (Friedman 2001, 77n6) Friedman suggests that there are benefits to using the more general definition via coordinating principles. The point for us is that the concept of

430

Scott Tanona

equal time is here defined theoretically via the a priori principles of a framework; we might therefore say that the observations are not merely theoryladen, but are framework-laden. The potential problem with such framework-ladenness is that the abstract framework cannot be both the framework constituting the framework-laden phenomena and the abstract formalism that only has empirical content through coordination with the independent empirical phenomena. Consider the light principle. Friedman claims that this principle is an empirical fact, something we can discover about the empirical world, and that in special relativity we elevate this fact to a coordinating principle. As a fact we learn prior to the establishment of relativity, it is discoverable through empirical means independent of the formalism of special relativity. Because it claims that the speed of light is the same in any inertial reference frame, the concepts of speed and light must be understood independently of the new formalism. Reichenbach himself recognized the possible problem posed by theory-ladenness and realized that if his concept of coordination was to work, he needed an independent theoretical grounding of empirical observations. Reichenbach's method of successive approximations (Reichenbach [1920] 1965, [1924] 1969) suggested that actual observations in relativity are laden with pre-relativistic physics but that they are legitimized as approximations of the more accurate relativistic physics. The version of Reichenbach's views most relevant here occurs in his 1924 Axiomatization of the theory afrelativity, where he suggests that the thing coordinated is neither an immediate perception nor direct reality but rather itself a conceptual construct: The difficulty concerning coordinative definitions is similar to that concerning elementary facts: the physical thing that is coordinated is not an immediate perceptual experience but must be constructed from such experience by means of an interpretation. If I establish the coordinative definition "a light ray is a straight line," then the coordinated physical thing, the light ray, is a construction going beyond perception. (Reichenbach [1924] 1969,8) We coordinate not direct perceptual experience, but rather theory-laden constructions. These constructions go beyond experience by interpreting perceptions in light of a particular theory. This interpretation is an expression of the theory-ladenness of measurement, of which Reichenbach gives a nice brief characterization: We shall have to make use of the scientific theory itself in order to interpret the indications of our measuring instruments. Thus we shall not

Theory, Coordination, and Empirical Meaning in Modern Physics 431 say 'a pointer is moving' but 'the electric current is increasing.' (Reichenbach [1924] 1969,5) This interpretation requirement is important because it generates precisely our difficulty: how can the concrete empirical phenomena be constructs within a theoretical fi-amework and also be independent from the theoretical framework? Reichenbach's answer to this difficulty is that the theories or fi-ameworks involved in the two parts of the question are not the same; we use the old pre-relativistic framework as an approximation upon which we can base, through successive approximations, the new theory: All axioms of our presentation have been chosen in such a way that they can be derived from the experiments by means of pre-relativistic optics and mechanics. All are facts that can be tested without the use of the theory of relativity. In particular, they are all formulated without the use of the concept of simultaneity at distant points ... The particular factual statements of the theory of relativity can all be grasped by means of pre-relativistic conceptions; only their combination within the conceptual system of a theory is new. (Reichenbach [1924] 1969, 6-7, emphasis original) Such a perspective would indeed address our problem, and I suggest that it is in part correct. However, the view I develop below differs significantly from Reichenbach's in a number of ways, primarily regarding the questions of whether such use of pre-relativistic physics might be avoidable and how it ought to affect our interpretation of the theory. 9

3. Layers of Frameworks One of the most interesting aspects of Friedman's relativized a priori is the idea that intersubjective and precise descriptions in physical theories can only be given in the context of a presumed conceptual fi-amework. However, I suggested above that if the concrete empirical phenomena themselves play such a significant role in specifYing the empirical meaning of our modern theories' abstract mathematical structures, then the phenomena themselves must be intersubjective and independent of the theories and their chosen fi-ameworks. My suggestion is that concrete phenomena do require their own conceptual fi-amework; the "same" phenomenon can be measured and specified in different ways, so it is not quite "concrete," but rather is conceptualized. However, the concrete phenomena can depend on a fi-amework different fi-om the one to which they are coordinated. Coordination would

432

Scott Tanona

then be a process of establishing a relationship between two different frameworks. I suggest that by amending Friedman's picture in this way we can both accept his fundamental insights and account for the independence that I have suggested is needed for coordination. These suggestions bear similarities to Bogen and Woodward's suggestion that theories make empirical predictions not about particular data points but rather about theory-laden phenomena such as the melting point of lead, or the existence of neutral currents (Bogen and Woodward 1988). On their view, a theory does not explain particular data points of an experiment (which are due to contingent and varied facts of the particulars of the situation) but only explains the experimental phenomenon. While I believe there may be some issues with their account (e.g., whether or not the theory can thereby also explain particular data points depends on one's idea of explanation), my account is not about explanation, and I do believe that often the best way to think about the empirical content of theories is in terms of their relation to entire phenomena. It is therefore worth considering how Bogen and Woodward's view applies to the account I am proposing. Bogen and Woodward's view is most easily understood in one of two ways. One way is to suggest that phenomena just are patterns in data, in which case they may be just as much a feature of the data points of a particular experiment as are the data points themselves.l° The other is to suggest that they are real features of the world, independent of whatever is done in a particular experiment or measurement. I suggest a third view: phenomena are theoretical constructs that can be specified only within a conceptual framework. Different frameworks can construe the "same phenomena" in quite different mathematical and theoretical ways. Here I borrow again from Duhem, who says that "the result of an experiment in physics is an abstract and symbolic judgment": What is the value of the volume occupied by the gas, what is the value of the pressure it supports, what is the degree of temperature to which it is brought? Are they three concrete objects? No, they are three abstract symbols which only physical theory connects to the facts really observed. (Duhem [1906] 1991, 146) The idea here is also what is dictated by Friedman's retention of the constitutiveness of the a priori-the objects of a theory are not concrete physical objects but framework-dependent representations of them. This applies just as well to the phenomena explained by a theory, as the objects of the phenomena are objects of some theory or other. This third view suggests that at least for certain physical theories, the en1pirical content comes through the theory's connection not just to a par-

Theory, Coordination, and Empirical Meaning in Modern Physics 433 ticular phenomenon, but rather to the entire conceptual framework of that phenomenon. l l As I will argue below, special relativity has empirical content through the connection between elements of Minkowski space-time and the classical space and time of an empirical inertial reference frame. Predictions about a particular experiment or measurement are then determined both by how the theory coordinates its formalism to the framework of the phenomenon as well as by what the theories of the framework of the phenomenon say about the particular situation. For example, special relativity dictates that the speed of light will be constant in any direction; classical, non-relativistic theories can then tell you how this will be evident in the particulars of, say, a Michelson-Morley setting, by the lack of changes in interference fringes (more on this below). Another example worth considering is the way Bohr's model of the atom fl.·om early quantum theory accounted for the atomic spectra.l 2 The model accounted for the Balmer series, which is a phenomenon concerning the patterns in the frequencies oflight emitted from excited hydrogen atoms. Because frequency is a wave concept, the predictions of Bohr's model therefore concerned properties of light according to a particular theoretical conception of light-the wave theory-and the conceptual framework necessary for the formulation of that theory. The model's empirical content depended on the way transitions between stationary states were associated with emitted frequencies of light. On this view, the Balmer series and other such series are the concrete empirical phenomena for the model. These phenomena are not simply directly observable but themselves are theoretical in nature. They are wave phenomena; they are regular patterns in the frequencies of light emitted by excited hydrogen atoms. They are phenomena only when conceptualized that way; if conceptualized another way, there may not be any such patterns at all (recall the choice between universal forces or a curved geometry). That is, their description and our understanding of them make sense only in a particular conceptual framework concerning the wave theory oflight. Because the model's empirical content concerns emitted frequencies of light, it assumes that framework for the phenomena. Thus any method of measuring frequency within that framework is an appropriate measurement of the atomic spectra accounted for by the model. In suggesting these ideas I am not merely describing the role background theories and auxiliary hypotheses play in deriving observations. The fact that one needs additional background theories to make predictions is consistent with the idea that the observations could be described in a pure observational language, which I deny. The point I am making is that the place where the background theories come in is important. On this view the observations are themselves theoretical in nature, and so one cannot make claims about these observations without invoking

434

Scott Tanona

the theories and foundations of those phenomena. In other words, the idea is not merely that we need auxiliary hypotheses and background theory to help us derive statements about observations. Rather, the background theory constitutes the empirical phenomena and establishes the theoretical language that describe them, and it is an assumption of the new theory that the observations are objects of the "background" theory. A new, abstract theory has empirical content via a coordination to this background theory and framework; it can thus make claims about the phenomena of that framework via the chosen coordinating principles. Now, even though coordination is between frameworks, the relationship is not symmetric. Rather, one framework is taken as a given for the other. The former is assumed to appropriately capture the aspects of the phenomena used as the empirical foundation for the other. Of course, that framework itselfwill not account for all the aspects of the phenomena (e.g., the Balmer series, or time dilation), but it does define the things involved in the phenomena (e.g., frequencies of light, distance and time) and thus serves as the theoretical space in which the empirical phenomena find their home. On this view some such conceptual frameworks are used to define the empirical content of others. What makes a framework empirical, then? I suggest that there is no real difference in kind between the two and that whether or not a framework is empirical is relative. However, ideally this use of one framework as the empirical foundation for another is justified in some way, perhaps because the one is somehow closer to "observable" in the perceptual sense. I will not try to give any real account of what makes a certain framework empirical for another here, although I will consider the particulars of the case for special relativity below. Now, the conceptual frameworks of this discussion, even the "empirical" ones, all involve abstract structures of geometry and laws of motion as well as dynan1ical theories. But these are merely abstract theoretical formalisms if they are not grounded to empirical phenomena. An empirical framework is just a conceptual framework being used to define thestructure of the phenomena for some other theoretical framework, so independent of this other one, the story will be the same as what is told for any-e.g., it requires coordination principles itself. That is why being empirical is relative. I call this view a framework -layered (or theory-layered) conception of empirical content: Connecting an abstract theory to actual observations involves a series of layers of theoretical frameworks. The relationship between theory and observations is mediated by other theories such that there are layers of theories or frameworks involved in connecting a theory to observations.1 3

The01')', Coordination, and Empirical Meaning in Modern Physics 435 I believe this conception of framework-layering to be potentially fruitful in physical theories in general. However, here I will focus mainly on its application to special relativity and then, below, to quantum mechanics.

4. Non-Relativistic Concepts in Special Relativity I argue that for special relativity to have empirical meaning, its phenomena must be understood within the context of their own, independent conceptual framework. Without an independent framework to indicate the empirical (and not the abstract structural-theoretic) meanings of light, distance, and time, the light principle would not be able to provide empirical content to relativistic space-time.l 4 The relativized a priori of special relativity is not all-encompassing and is not constitutive of either the empirical phenomena or our experiences of them. Both of these occur from the perspective of a particular frame of reference, and within that perspective we have an independent notion of space and time, what it means to make a measurement of the speed of light, and so on, independent of relativity. I therefore suggest that the form of Minkowski space-time does not concern the form and content of our experience and observations within a reference frame. From the perspective of special relativity, the objects of physics (the theoretical and mathematical abstract objects of the theory) exist in the framework of relativistic space and time; the objects of "experience," which constitute our observations (relatively speaking, given the above caveats), exist in a pre-defined, pre-relativistic framework. What is this independent framework? Consider first the way we established the empirical fact that the speed of light is constant. Although the Michelson-Morley experiment originally was meant to measure the speed of earth relative to the ether, we now treat it as having established the constancy of the speed oflight. The speed oflight was thought to be constant in the ether frame, and the earth had to be moving relative to the ether in one direction or another. To try to measure the speed of the ether wind relative to the earth, Michelson and Morley set up an apparatus for measuring differences in the speed of light in two different directions. The apparatus split a beam of light and sent it along two perpendicular arms with mirrors at each end, so that the beams were reflected back and then recombined. Because of the presumed motion of the earth relative to the ether, there would be some orientation of the apparatus (which could rotate) in which light along one arm would be traveling with and then against the ether "wind" while the other would be traveling perpendicular to that wind. In this orientation, light traveling along the first arm would be expected to take longer to travel the round-trip than would light on the second, and the difference in the time it would take for the light to travel

436

Scott Tanona

the distances along the two directions would show up as a phase difference measurable via interferometry. If the apparatus were rotated, then at some orientation the apparatus would be in this state of maximum difference between the time to make the one round-trip and the time to make the other. At some other orientation there would be no difference between the times. Michelson and Morley looked for changes in the interference pattern as the apparatus rotated, and found no such change. This lack of change in the interference pattern indicated that the speed of light along the two arms did not change when the orientation changed, and so indicated that the speed of light is the same in any direction in the lab fran1e. 15 This empirical phenomenon poses a problem for the ether theory if one uses the classical Galilean transformation rules that translate the velocity in the ether frame to the apparent velocity in the reference fran1e moving relative to the ether by adding or subtracting velocities. On the basis of these rules, the velocity of light relative to the lab frame ought to equal its velocity in the ether minus the velocity of the earth relative to the ether. Maxwellian ether theory says that the speed of light is c within the Maxwellian ether rest frame. Under the assumption that the Earth is moving relative to the ether in some direction, the speeds as measured in our reference fran1e ought not be the same in most orientations of the apparatus in the lab frame, and there would be some difference in the interference pattern as the apparatus turned. But the empirical phenomenon is that the speeds are the same. Special relativity accounts for this phenomenon by dropping the Galilean velocity addition rules and introducing the Lorentz transformations for calculating frame-dependent distances and time in different inertial frames. This change establishes by stipulation the validity of Maxwell's equations within any frame (not merely the ether frame of rest). It also drastically alters our conceptions of space and time outside our reference frame. The Lorentz transformations indicate that the distance and time between two events as measured in one inertial reference frame will in general differ from the distance and time measured in a different one. According to special relativity, there is no global time and no global space because there is no global simultaneity relation. Instead, we have a global four-dimensional manifold of space-time events upon which our referenceframe-dependent simultaneity relations are projected. There is a welldefined and invariant space-time interval between any two events, and it is, for example, an invariant feature ofspace-time whether lines between those events are time-like, space-like, or null. But the distance or time between two events is reference-frame-dependent. As a consequence, certain features of relativistic space-time become difficult to interpret; in particular, they are difficult to understand in terms of common intuitive ideas about

Theory) Coordination) and Empirical Meaning in Modern Physics 437 space and time (for instance, those concerning whether the future is undetennined) because space and time have no well-defined global meaning. Because special relativity tells us that the world is not really as we thought, one might question the degree to which the independent concepts of space and time apply at all to the world according to relativity. However, despite the fact that the concepts of space and time do not apply globally, they remain valid concepts within an inertial reference frame. Space and time have not been dropped in relativity. There remain robust notions of "proper" time for and "proper" length of an object that correspond to the time and space of the inertial reference frame in which it is at rest; indeed, the proper length can in some respects be understood as the "real" length of an object. I argue not only that the empirical content of special relativity relies on our having independent conceptions of space and time that apply within an empirical reference frame but also that special relativity takes them for granted and does not explain what they are or why they apply.16 One of the most important things we hold onto in these empirical descriptions is the distinction between space and time. Indeed, anyempirical phenomenon in a reference frame is a phenomenon in Euclidean threedimensional space plus time. Of course, special relativity tells us that this space and time is only a particular decomposition of space-time into individual space and time components. Space-time itself has no global simultaneity that would allow us to apply this particular perspective globally. Nevertheless, the phenomena of the reference frame do occur with a particular such perspective. Moreover, it seems we cannot avoid this situation. First, we arguably have no way of measuring the space-time interval between two events without first measuring the distance and time between those events with respect to a particular reference frame. All our knowledge of the relationships between space-time events comes via a particular space-time perspective. Second, the nature of relativity is that its empirical content (its predictions, etc.) concerns precisely the individual space and time perspectives of different reference frames. Indeed, that is exactly the point of Lorentz transformations. Special relativity assumes that all the objects at rest in an inertial reference frame share the same common perspective of space and time. It can justify this assumption to some degree, as the Lorentz transformations return the same values of distance and time for two objects at rest with respect to one another. Yet it is also part of the fundamental assumptions of a reference frame that this is true. Moreover, there is more to a reference frame than just the objects at rest with respect to one another. The novelty of relativity concerns what it tells us about the application of the reference-frame-dependent perspective of space and time to moving

438

Scott Tanona

objects. An object moving relative to that reference frame "experiences" a different perspective of space and time than an object at rest in that reference frame. But the space and time of the objects at rest in that frame also apply to the n10ving object with respect to that frame, and the Lorentz transformations tell us how observers in motion with respect to one another disagree on simultaneity, elapsed times between events, and the lengths of objects both at rest and in motion with respect to those frames. However, relativity does not tell us why the space and time of the objects at rest apply to objects moving in that frame as well-it just assumes that they do. Part of what it means to be a reference frame (inertial or not) is that it establishes distinct space and time dimensions that are applied to all objects, moving or not, from the perspective of that frame. Relativity just tells us how the distinct space and time perspectives of different inertial reference frames are related.

5. Non-Relativistic Framework$ in Reference Frames What does it mean for there to be a distinct space and time of a reference frame but not of the global world? That is one of the interesting questions to be asked in the interpretation of relativity, and one without a straightforward answer. While relativity gives a mathematical account of these relations, it is less clear what ontological account we can give of them. Central to the topic is the nature of inertial reference frames. In Minkowski spacetime, an inertial reference frame is associated with a coordinatization of all space-time, so at least in that abstract representation, it is infinitely extended. But reference fran1es are also concrete and empirical, picked out by actual, concrete objects that serve as reference points. As such they are tied to something much more local in nature-not necessarily a set of rods and clocks, but something like an actual laboratory, a train or rocket ship, or even an entire planet. We therefore need to distinguish a reference fran1e itself from both the concrete objects with respect to which that frame is established and the framework of that frame (and, in particular, the coordinate system of that frame). Moreover, we need to distinguish all of these from a particular coordinatization of Minkowski space-time which we may take to correspond to them. I am suggesting that the abstract structure of Minkowski space-tin1e really only applies to empirical space and time via the coordination between it and the independent fran1ework of empirical phenomena. This occurs through the establishment of a correspondence between substructures of Minkowski space-time and reference frames. Note that these reference frames are not merely the coordinatizations in Minkowski space-time, as it

Theory, Coordination, and Empirical Meaning in Modern Physics 439 is the latter to which the former are coordinated. So what are the former? The reference frame is a particular point of view, picked out by a particular object. In that sense, it may be concrete and unconceptualized, but in its role of providing for the measurement of position and time, it becomes an independent conceptualization of the wayan object establishes a perspective of space and time with respect to that object, and includes therefore not only the geometrical conception of classical space and time of a frame (i.e., relative to an object) but also the physics required for establishing measurements of those things. A coordinatization in Minkowski space-time does not provide all these. Special relativity does not define what space and time are in a reference frame. The light principle may on some views provide the canonical measure of distance and time of distant events-we can send a signal out and back, assume the speed is the same in both directions so that the event of hitting the other object occurs halfWay through the process, and determine the distance to the object at that time based on the speed oflight. 17 Yet it can serve this role only with an already presumed notion of time. What it means to say that the light principle coordinates abstract structure with empirical phenomena is that that measure defined through this procedure is a measure of distance, so that all our other ways of measuring distance, etc., correspond to that one. Thus the light principle tells us how to correctly apply (or translate) our independent notions of space and time, especially with respect to moving bodies. The space and time of a reference frame are then associated with the theory when the coordinates of the reference frame are put into correspondence with a particular coordinatization of Minkowski space-time. The difficulty with taking special relativity to define a new notion of distance can be seen by considering the question of how we fix a notion of inertial reference frames. One way of viewing relativity is that inertial frames are now defined by the light principle. Since the light principle dictates that the speed of light is constant in any inertial reference frame, it tells us that any reference frame in which the speed of light is not constant is not inertial. But the constancy of the speed of light can only fix a notion of inertial reference frame if we have prior notions of space and time, and, in particular, a prior understanding of reference frames, so we must have an independent conception of and ability to apply space-time notions. IS Note then that if the light principle is used to define distance, it can no longer define inertial frames. On this view, the light principle tells us which reference frames are inertial, and then we can associate each of their spaceplus-time frameworks with a coordinatization of Minkowski space-time. I suggest that this prior, independent account of space and time in a reference frame is carried forward from our pre-relativistic conceptions. Consider the theories involved in the Michelson-Morley experiment and

440

Scott Tanona

the experiment's interpreted "null" result. While Michelson and Morley certainly had direct perceptual observation experiences while performing their experiment, their results did not concern those particular, direct experiences. Rather, those direct, phenomenal experiences were interpreted as "observations" concerning the fringes in the interference pattern and ultimately to the "observation"-the empirical phenomenon-that the speed of light is the same in any direction in the reference frame of the lab, stationary on the earth. I suggest that it is these "observations" that we think of as the experimental result, and the generalization of these observations that the speed of light is independent of an observer's motion is what is elevated to a coordinating principle on Friedman's account. Michelson and Morley assumed a three-dimensional Euclidean space and a distinct tin1e dimension, as well as good chunks of wave theory and plenty of other classical physical conceptions. These theoretical tools together form a framework in Friedman's sense: there is a chosen geometry of space and time, a Newtonian kinematics on top of the geometry, particular theories of light propagation based on these other foundations, and so on. Michelson and Morley's result that the speed of light is constant cannot be inferred and does not even have its intended meaning outside the conceptual framework in which speed, light, etc., are all understood in classical and wave-theoretic terms. To work as a coordinating principle, the light principle must retain the pre-relativistic framework for the description of empirical phenomena. I suggest, then, that even after the development of relativity, we cannot abandon the pre-relativistic framework. One could replicate the Michelson-Morley experiment today and use entirely the same theoretical inferences to interpret the experiment. Of course, the point of the experiment would not be the same, since with relativity the speed of light is postulated to be the same. If the experiment turned out otherwise, one would conclude either that there was some mistal