867-POSC9-4.vp The Dilemma of Case Studies: Toward a Heraclitian Philosophy of Science Joseph C. Pitt Virginia Tech What do appeals to case studies accomplish? Consider the dilemma: On the one hand, if the case is selected because it exempliªes the philosophical point, then it is not clear that the historical data hasn’t been manipulated to ªt the point. On the other hand, if one starts with a case study, it is not clear where to go from there—for it is unreasonable to generalize from one case or even two or three. After Kuhn cast doubt on the usefulness of abstract positivist models by appealing to the history of science,many philosophers have felt compelled to use historical case studies in their analyses.Kuhn, however, did not tell us how todo this. Further, it is not clear exactlywhat appeals to case stud- ies accomplish.We can frame this issue as a dilemma.On the one hand, if the case is selected because it exempliªes the philosophical point being ar- ticulated, then it is not clear that the philosophical claims have been sup- ported, because it could be argued that the historical data was manipu- lated to ªt the point. On the other hand, if one starts with a case study, it is not clear where to go from there—for it is unreasonable to generalize from one case or even two or three. I will argue that even very good case studies do no philosophical work. They are at best heuristics. At worst, they give the false impression that history is on our side, sort of the history and philosophy of science version ofManifest Destiny. If historical studies are to be useful for philosophical purposes, theymust be extended historical studies that contend with the life span of a scientiªc problematic. It is not enough to isolate a single ex- periment or to look at the activity of a lab under one director. One needs to place the case in the context of a problematic and to explain a problem- atic in terms of its origins and its fate (Pitt 1992). But even if thiswere to be accomplished, it is not clear what philosophical work is being done. 373 Perspectives on Science 2001, vol. 9, no. 4 ©2002 byTheMassachusetts Institute of Technology borrego Typewritten Text Copyright by the MIT Press. Pitt, Joseph C. "The Dilemma of Case Studies: Toward a Heraclitian Philosophy of Science," Perspectives on Science, Winter 2001, Vol. 9, No. 4, Pages 373-382 doi:10.1162/106361401760375785 This may be, at best, history of ideas. The point here is simple: just as philosophical problems are not problems about the single case, historical issues are particular andmust be seen in context. But seeing an historical issue in context does not by itself suggest any particular philosophical point. It may be that the problem here lies in our understanding, or lack of it, of what constitutes a context. The importance of understanding the appeal to historical contexts is to show how doing history in context lim- its the possible range of philosophical ideas and explanations. By way of example, I will consider the philosophical question of what constitutes a scientiªc observation. I will argue that a serviceable universal account of scientiªc observation is not possible because the activity of making a scientiªc observation depends on, among other things, the sophistication of the technology available at the time, hencewhatwemeanby a scientiªc observation changes. What is allowed as an observation varies in time, place andwith respect to changing criteria inºuenced by technological in- novation. If I am right, this view provides a serious basis for rejecting Kuhnian paradigms. Problematics have histories, but that does not mean they are stable over time.Quite the contrary, the reasonwhy it is important to ap- peal to problematics is that they change even as they serve to restrict re- search to certain topics. And it justmay be thisHeraclitian characteristic is a deªning feature of science.As philosopherswe seek universals, but the only universal regarding science is change. That seems to be a fact. But, it might be responded, as philosophers we are also interested in the norma- tive—our job is to attempt to show what we ought to mean by x or y. While that is true, in our normative guise we also cannot ignore what in fact is the case. The hard job is to ªgure out how to do that. The lesson to be learned is that if philosophers wish to use historical cases to bolster their positions, thenwewill have touse very long studies andwewill have to ªgure out how to relate the history to the philosophical point without begging the question. The issue of not begging the question looms large. Let us start with a big question, which is continually begged: just what constitutes a case study? This goes to both horns of the dilemma, but particularly to the question of how to avoid appearing to manipulate historical data to ªt philosophical theories. Forwithout credible criteria for selecting or identi- fying a case as a case the charge can be legitimate. Despite the currency of case studies, there are currently no criteria available to ascertainwhenwe have one before us.We select the historical episodes we do for a variety of reasons with few, if any, operative guiding principles. I propose that we can develop a set of criteria for selecting a case study, but there are several costs. The problems involve the selection 374 The Dilemma of Case Studies criteria. For example, if we want to start with the science and see where that leads us, then,without begging the question,we have to ªnd the sci- ence. Identifying the science in question in a non-whiggish fashion is a delicatematter.We simply cannot assume thatwhatwe call physics today, is what the scientists practicing physics in 1830 would call physics. We can’t ªnd the case study because we can’t ªnd the science in which it is a case. But there is a way out. Theway out is to proceed by identifying a problematic. A problematic consists of a set of intellectual concerns thatmotivate a scientist or a group of scientists to pursue the investigations they do. I suggest that this char- acterization skirts the demarcation question because where a group of in- vestigators canbe identiªedwehave a social fact as a startingplace. For an example of such a group, I suggest Copernicus, Tycho, Kepler, Galileo, Clavius and Scheiner. Their interests need not constitute a one-one correspondence, but each had to consider what the others had to say as relevant to their research interests, either singularly or in sets. Now for the cost: problematics have their own history, they have start- ing points and end points, and in between they change, mutate, some- times they evaporate, sometimes theymetamorphise into something new. Further, in the course of working within the problematic, what emerges may not be what was expected. Finally, although this may seem obvious, to identify a problematic one must position it historically. This is to put the problematic in context, which is difªcult, for in any historical setting there are many contexts, and we must avoid begging the question by se- lecting a contextwhich conveniently supports our concerns. In short, ifwe start with case studies, we are assaulted on all sides by issues of question begging. Let us lookmore deeply at the notion of context. Whatdophilosophers expect to accomplishby appealing tohistory and historical contexts? We all know that, contrary to popular belief, Kuhn was not the ªrst to wag an historical ªnger at us; Norwood Russell Hanson was doing history and philosophy of science in the ªfties and his work was well received within the inner circle. Lakatos, borrowing freely fromKant, asserted that philosophy of sciencewithout historywas empty and history of sciencewithout philosophywas blind. Inwhat sense is phi- losophy of science without history blind? Have we not been able to see clearly through the lens of logic to important structural characteristics of, for example, explanation and conªrmation? If the claim is that what we have comeupwithdoesn’tmatchwhat scientists actually do, then it is not clear that that is a valid criticism sincewe have a normative, notmerely a descriptive role to play. Determining the logic of key concepts andwork- ing that out is a perfectly legitimate activity. What is it that history is supposed to supply? Perspectives on Science 375 In part, this is a question concerningwhat it is we thinkwe are doing. Or to be brutally frank, what are the goals of philosophers of science who use history in someway or other, ormore speciªcally,who see the appeal to historical context as important? Let us begin by reviewing the evils contextualization is supposed to avoid: 1. WhigHistory; a term coined byHerbert Butterªeld (1931), it re- fers to the attempt to impose current categories of analysis on his- torical events. 2. Universalism—a corollary to (1); the idea that certain features of science are constant over time. 3. Modernism; the insistence that the most important developments of any epoch are science (conceived in contemporary terms) related. 4. Abstraction; the reiªcation of key features of a period. 5. Internalism; the process of examining the work of a person by ap- peal only to his or her notes and texts without consideration of any social or external factors—falls prey to (3) or to (1), since to really know is already to understand the context in which an author writes. Assuming it is possible to avoid the above, there remain serious dangers facing the contextualist. For what the historian concentrating on context does, having avoided these ªve cardinal sins, is to concentrate on individu- als and to consider the inºuences on and the consequences of these inºuences for their work. What this means is the following. First, given (5), all that the historian can do is to reveal the social and intellectual fac- tors that might be said to motivate the views expressed by the particular historical ªgure under consideration. For to provide a close analysis of the work of the person in question (itmust be a person to avoid (3) or (1)), ex- posing its logic or even its content amounts to (5). To the extent that the views of some person or other are to be considered, it is only by virtue of his perceived audience or inºuences. But determining who are the audi- ences and inºuences falls to the historian to identify since, we are told, historical ªgures cannot be trusted to know whom they really are inºuenced by or to whom they are really responding. How the historian avoids (1), (3) or (5)—or howhe or she knowswhomto identify as the rel- evant audiences or inºuences—remains something of amystery.Theprob- lem here is fundamental: (a) it is not enough to say ‘x read y’—since that alone does not establish inºuence except in a trivial manner; (b) nor is it enoughthat xquotedy or thatxadmitted to either reacting to y or even at- tempting to extend y’s position—since x may not knowwhat reallymoti- vated him or her. (The contextualist has opened the door to this objection 376 The Dilemma of Case Studies byusing it to reject Internalism, i.e., tu quoque); (c) it is equally inadequate to cite who read x, for it can be the case that x was read for all the wrong reasons—(c.f., themisuse ofNietzsche by the 3rd Reich). The contextualist historian is now left in the position of arbitrarily identifying people in places and can only hope that the preponderance of the evidence and correlations account for what x said about y. The laud- able intent of the contextualist is to show that great ªgures do not emerge from a vacuum. The problem, however, is that there is no obvious princi- ple of selection that guides the identiªcation of people who or events that allegedly transform the vacuum into a social context. The result can be that the ªgures highlighted can be minor or obscure; likewise for social factors. Without a well-articulated and defensible principle of selection, the attempt to construct a context is at best arbitrary; at worst it is self-serving. Why certain ªgures are identiªed is also not clear, since all the objections used above with respect to x apply equally well to these problems. The contextualist project, seen in this light, is hopelessly ºawed. Aswe have seen, if we pay too close attention to the standard justiªcat- ion for contextualization, the program collapses. And yet there is some- thing positive to be said in favor of each of (1)-(5), i.e., the rejection of Whig History, Universalism, Modernism, Abstraction, and Internalism. It is just that taken together nothingmuch is left.Havewe taken awrong turn somewhere? It might appear that we have been led to our unhappy conclusion by concentrating on only one aspect of the contextualization of history, i.e., individuals. But the collapse of contextualism does not occur only when individuals are the subjects of discussion. For example, an anti-Whig his- torianwill also justiªably reject talk of “science” in the 16th century, there being natural philosophy for the study of the natural world. Thus, the reiªcation of concepts also seems to be a problem. So, what is the point of contextualization? What is the appeal to con- text supposed to accomplish?Minimally a context is supposed to provide an explanatory framework for speciªc historical developments, i.e., it sets the stage onwhich the historian’s explanations will be seen tomake sense when offered. The crucial mistake made by advocates of historical contextualization is to give the impression that there is only one appropri- ate context that satisªes the explanatory-allowing role. Thewriting of his- tory is necessarily selective. However, the shift from individuals or activi- ties such as history or art to context is no less selective or arbitrary, for (with apologies to Nelson Goodman) contexts are where you ªnd them. For example, consider the contexts inwhichGalileo could be said to have operated. (1) The Renaissance, (2) The Scientiªc Revolution, (3) The Perspectives on Science 377 Medicean Court (pace Biagioli), (4) The Archimedean tradition, (5) The Euclidean tradition, (6)TheAristotelian tradition, (7)ThePlatonicTradi- tion, (8) The Medieval tradition, (9) the battle between the Vatican and the Italian secular states for political control of the Italian peninsula, (10) the Age of Exploration, (11) The Age of Elizabeth, (12) the Sixteenth Century, (13) the Seventeenth Century, (14) a personal struggle to ªnancially support his family, (15) the personal politics of the struggle be- tween theologians and natural philosophers (pace Redondi), (16) The Counter Reformation (pace Shea). And so far we haven’t even begun to ex- plore whether we should approach Galileo as an engineer, a physicist, an astronomer, an instrumentmaker, an amateurmusician, a father, a philos- opher, a theologian, a good catholic or an irritation of the Pope’s. However, picking the relevant explanatory framework may not be as difªcult as I appear to be suggesting. The trick lies in ªguring outwhat it is about the person or the event you want to explain. The mistake to be avoided is assuming there is necessarily only one explanatory framework. Even so, there is something more problematic than determining which framework topick, that is the problemof determiningwhat constitutes an appropriate explanatory framework or frameworks for a topic, i.e., what constitutes an explanation in these contexts, orwhat constitutes an histor- ical explanation simpliciter. To ask this question assumes that there is one kind of historical expla- nation that ªts all sizes. Clearly, this is not the case.We actually have two questions here—ªrst, there is the problem of selecting an appropriate framework. Second, once a framework has been selected, we still need to be able to sort outwhat kinds of explanations are appropriate and satisfac- tory andwhich ones are not. Answering these two questions is clearly be- yond the scope of this paper. Iwill concentrate here only on onepart of the second question and I will do so by trying to answer a slightly different question, namely, “What do we want from an historical explana- tion?”—i.e., what is the point? Rephrasing it, the question reads, “Why dowe seek historical explana- tions,” which sounds a lot like, “Why do philosophers of science turn to history?” One tried and true answer is, “To learn from the past.” It is un- likely, however, that we seek historical explanation only to understand howwegot towherewe are now.We seekmore fromhistory—notmerely an answer to the question, “Howdidwe get here?” but also, “How canwe avoid ending up in this situation in the future?” There is little doubt that that question cannot be answered for several reasons. First, the analogies between the past and the present are just those, analogies. Learning from the past is only as successful as the strength of the analogy between past and present, and in drawing the analogy we need to be careful not to fall 378 The Dilemma of Case Studies into the trap of doingWhig history. Second, there is no single fact of the matter of the past—more information is constantly surfacing, depending on what we think we need to know. Ideologies, cultural fads, etc., also inºuence the plasticity of our histories. And yet the situation is not hopeless. The search for criteria by which to select frameworks to use in obtaining answers from the past depends as much on the perceived state of the present as on our perception of the op- tions for the future.And it is in the latter thatwewill ªndout clues to the adequacy of historical explanation.The central idea is the notion of a coher- ent story. Whatmakes for an adequate explanation is the sense that our ac- count of why things happened in the past hangs together with what we knowproceeded and followed the event in question. Appearances to the contrary notwithstanding, I want to argue that this is notWhiggish. Nothing in this suggestion of a coherent story suggests that we necessarilymust see what happened in the past in a direct, causal line with the future, which is our present. It therefore makes no sense to talk about the global importance of current events, theories, etc., since that verdict awaits the future. A relevant set of contexts can be identiªed in terms of their explanatory value, i.e., the coherence they contribute to the story accounting forwhywhat happened happened.One is justiªed in expanding the set of contexts to the extent that the failure to include cer- tain factors can be shown to be relevant to understanding what happened after the events in question. So, an historical context is a set of factors that provide an explanatory framework for an event, a person’s actions orwork, or a social trend, etc. The adequacy of the context is a function of its abil- ity not only to account for the event in question, but also for its prior and subsequent history. All that havingbeen said,we still cannot account for the philosopher of science’s appeal to history. The job of explainingwhy the pastwas the past is the historian’s job. The philosopher who looks to the past as revelatory of the present is doing bad history, so that can’t be the justiªcation.Nev- ertheless, there is a philosophical job to be done with respect to the past. One of the features that needuncoveringwhenwe try to understand an in- dividual’s actions is the set of assumptions with which he or she is work- ing. In particular, we need to knowwhat were the expectations at play at the time in order to assess the quality of theworkbeingdone.Uncovering assumptions and exploring texts for hints to expectations are jobs philoso- phers are good at. But in so doing, we learn little about what is relevant for today. So, at the moment, it is not at all clear what the cash value of case studies is for the philosopher of sciencewho starts with history. Let us now turn to the other horn of the dilemma. Instead of starting with historical cases selected for the way they are assumed to illuminate Perspectives on Science 379 contemporary philosophical issues, or for providing the data for building a philosophical theory, let us start from the side of theoretical questions. Unfortunately, things don’t get any better. The kind of question I have in mind is, “What is a scientiªc explanation” or, “What is a scientiªc obser- vation”—when we look to history to answer such questions, we stumble inmanyways over assumptions that at ªrst seem innocent and yet eventu- ally prove fatal. For example, when one asks what is a scientiªc observa- tion, it seems that we are asking about the “observation” part, assuming that we know what “scientiªc” means. But even if we have a good solid understanding of what “scientiªc”means (which we don’t), we can’t sim- ply assume that we can apply that understanding backwards in time— to do so is to engage in Whig historiography, which we all now know is inappropriate. Now, let’s assume that we not only knowwhat “scientiªc” means, but also what “observation” means and what “scientiªc observation” means (which we don’t)—now each of these expressions has a history and their meanings have changed over time. To look toGalileo’s telescopic observa- tions to enlighten us as to themeaning of “scientiªc observation” today is to run rough shodover goodhistoriography and to assume that philosoph- ical analysis has some sort of a temporal a priori intellectual legitimacy and that as philosophers we can appropriate history to our own ends, conªrming our assumptions. What would it mean for Galileo to make a scientiªc observation of themoons of Jupiter? “Scientiªc” is not a term in play at the time. To claim that his observations were scientiªc is to read backwards from the present into history, which is unjustiªed. Second, it is not clear that at the beginning of the seventeenth century there was a formal understanding of what wasmeant by an observation as opposed to anynumber of other similar activities such as seeing, perceiving, sighting, etc. Finally, with the advent of new instruments we can trace the transfor- mation of the concept of an observation. We can agree on why certain highly constrained settings in a lab can yield observations.Butwhat about the pictures of the surface of Io being sent back from the Galileo probe? There are a number of different kinds of steps in between the taking of a measurement of Io and our seeing the result here on earth. Transmitting devices need to be aligned, involving computers and computer programs. There is the encoding of the measurement and then the sending and the assumption that nothinghappens to itwhile itmakes itsway from the or- bit of Jupiter to Earth. Then there is the reception of the data,more com- puters and programs to transform the encoded data into a picture and Lo! An observation! To accept those pictures as observations requires an ex- panded understanding of what constitutes an observation from the simple 380 The Dilemma of Case Studies naked eye seeing of nature and our report of that seeing to something con- siderably more complicated and sophisticated. The extent to which we have accepted the fact that we can use remote instruments tomake obser- vations is a far cry from simple seeing. I propose that not just observation, but all of the concepts we use to discuss science are in constant ºux. PeterGalisonmakes that casewith re- spect to the meaning of “experiment” in the 20th century. What consti- tutes an explanation, evidence, data, observation, etc., all change over time and usually in response to some technological innovation.1 That be- ing the case—i.e., that the meanings of these concepts are in constant ºux—it would seem impossible that we could learn anything about our present concerns fromthepast.And so once again, the question remains as towhatwe can gather from case studies. Sowhere does this leave us?Wedon’t knowwhat a case study is—ifwe shift to a problematic we open up a can of worms—problems are embed- ded inhistorical contexts, but selecting the right contextwithout begging the question isn’t obvious. On the other hand, if we assume that concepts associated with philosophical analyses of science have some sort of atem- porality, we violate legitimate historiographical concerns. Does thismean thatKuhn’s wake-up call to philosophers to pay atten- tion to history was misguided? I think not. However, as philosophers we need to lower our sights, or perhaps we need to raise them and consider more than only abstract philosophical criteria. Further,weneed todevelop a more robust sense of the sloppiness of our conceptual history. We seek precision, deªnitional clarity, analytic sophistication. These are good— but there is more to understanding: depth, ºexibility, and a sense of the give and take and contingency found in history.2 References Biagioli, Mario. 1993. Galileo Courtier: the practice of science in the culture of absolutism. Chicago:University of Chicago Press. Butterªeld, Herbert. 1931. The Whig Interpretation of History. London: G. Bell. Galison, Peter. 1998. Image and Logic: a material culture of microphysics. Chi- cago:University of Chicago Press. Goodman, Nelson. 1953. Fact, Fiction, and Forecast. Cambridge, Mass.: HarvardUniversity Press. Perspectives on Science 381 1. Documenting this claim is the object of a project currently in progress, Seeing Near and Far, a Hericlitian Philosophy of Science. 2. This leads me to believe that my colleague Richard Hirsh may be correct when he suggests that if you can’t call the guy up and interview him it isn’t history. Hanson,N.R. 1961. Patterns of Discovery: an inquiry into the conceptual foun- dations of science. Cambridge: CambridgeUniversity Press. Kuhn,Thomas S. 1962. The Structure of Scientiªc Revolutions. Chicago:Uni- versity of Chicago Press. Lakatos, I., 1971. “History of Science and Its Rational Reconstructions.” In The Methodology of Scientiªc Research Programmes. Philosophical Papers Volume I. Editedby JohnWorral andGregoryCurrie.Cambridge:Cam- bridgeUniversity Press. Pitt, J.C. 1992. “Problematics in the History of Philosophy.” Synthese, Vol. 92,No. 1 ———. “Seeing Nature: origins of scientiªc observation.” In Conceptions de la Science: Hier, Aujourd’Hui et Demain. Edited by R. M. Burian and J.Gayon. Forthcoming. Redondi, Pietro. 1987. Galileo Heretic. Princeton,N.J.: PrincetonUniver- sity Press. Shea, William. 1972. Galileo’s Intellectual Revolution: middle period 1610–1632. NewYork: ScienceHistory Publications. 382 The Dilemma of Case Studies