.3. ,c a. W um." n... 3.31: "a? . ll a. . . .. .. five: . 2i; :1... 5.79.311: . .1}: ‘5... l: .i. . .x .3 I. 5.1.13.3... ‘ . I... 2.2.5.7....- .. 2...... s .4529 .7 l; x. . . . . , 9.... h... I! re.‘ {:25}. . . (93.119: 17.3 .3... . .1 t a... 3x 7 . hrs“... 1. 4.. p . ‘ I 3 . L.\ : . . 3.; 3.5. 1. $131. ‘1 1. “mm: crfi the.“ 5.5:... {3 - is... if“. ‘1. .1: i. uUI.‘ 5|. r I ma i 1.5.1: {:3 ‘ v .I ilii». L.l..1.uk.\|..f 7113' 31‘ it? .3... Hi}: 5;...bnit u... .l...‘ u: ¢2V"‘VO‘(1 THESIF‘ 3 2009 This is to certify that the thesis entitled TOWARDS A PRAGMATIC THEORY OF MEDICAL EXPLANATIONS OF DISEASE presented by Barry DeCoster has been accepted towards fulfillment of the requirements for the PhD. degree in Philosophy 324 fl,%/ Major Professor’s Signature 7/2 c / 0‘ Date MSU is an Affinmtive Action/Equal Opportumy ”1811le m ' LiBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 2/05 p:/ClRC/DaleDue.indd-p.1 TOWARDS A PRAGMATIC THEORY OF MEDICAL EXPLANATIONS OF DISEASE By Barry DeCoster A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Philosophy 2006 Willi C0 illness to patient dissertation. I It! march. have b sicknesses. largl explanations of Engaging palien generating info mike autonomol “’6 Would like It l‘tlalli‘lsm‘ and i ABSTRACT TOWARDS A PRAGMATIC THEORY OF MEDICAL EXPLANATIONS OF DISEASE By Barry DeCoster What counts as a “good” medical explanation? How should clinicians explain illness to patients? Can only medical clinicians generate these explanations? In my dissertation, I argue that explanations of disease, as derived from philosophy of science research, have been applied wrongly to the clinical practice of explaining patients’ sicknesses. I argue for an improved theory of pragmatic medical explanations where explanations of health and disease are collaborative efforts between doctors and patients. Engaging patients as epistemic agents of explanation empowers them as participants in generating information regarding their own health, thus improving patients’ abilities to make autonomous and informed health care decisions. Such an approach meets with all we would like to say about scientific explanations, i.e., that they do not devolve into relativism, and they retain scientific meaning. I provide an overview of the major theories in philosophy of science about scientific explanation in Chapter 2, including the works of Hempel, Salmon, Kitcher, van Fraassen, Schaffner, and Railton. Paul Thagard (How Scientists Explain Disease) and Kenneth Schaffner (Discovery and Explanation in Biology and Medicine) have argued medical explanations should be structured as scientific explanations. In Chapter 3, I critique this move as improperly characterizing the relationship between two types of medical explanations: biomedical research explanations (from the laboratory) and clinical explanations (developed as part of doctor-patient interactions). Philosophers of science have advocated primarily antic models of disease explanation that focus on complex causal-mechanistic interactions, e.g., between the environment, the body, genetics, and infectious agents. Yet ontic approaches remain inadequate as a basis for clinical explanations given that such approaches to medical explanations often fail to meet patients’ explanatory needs. I argue in Chapter 4 that clinicians and patients generate medical explanations through collaborative efforts. I utilize Lynn Hankinson Nelson’s concept of communities (rather than individuals) as primary epistemic agents. I show how clinicians, patients, and clinician—patient groups shape explanatory requests and responses. This rectifies Thagard’s oversight by recognizing how willing and competent patients often take on active and beneficial roles in generating medical explanations. Rather than rely upon exclusively causal-mechanistic schemas, I argue in Chapter 5 that medical explanations are best understood as erotetic projects (explanations that answer the situated why—questions of the particular patients and clinicians involved). I draw heavily upon Bas van Fraassen’s work on pragmatic explanations in The Scientific Image. This model allows the generating of medical explanations to be collaborative projects that account for both clinicians’ and patients’ interests as shaping medical explanations. Their personal interests shape the why-question asked, as well as the contrast classes and relevance relations (that is, the explanatory concepts that regulate what information appropriately can be considered as part of the medical explanation). In Chapter 6 I argue for the inclusion of the values of a feminist science (as developed by Helen Longino) as additional pragmatic features that shape medical explanations. This move allows for articulation of the moral-political components of generating and evaluating medical explanations, explanations as projects undertaken to reflect the interests of clinicians and patients. These improvements to medical explanations allow for a wider range of explanatory work, for example, accounting for social determinants of health as explanatory. I “Quit thank Fred Gif project. I also I Ante. and Let I also ti. Hmanities in t Gaiton Thom to teaching has “Grid and to [iii hm tangm me . M} g: Wolf. Jennifer [Cachm‘ and . can be named Pm of [his di. ACKNOWLEDGMENTS I would like to thank a number of people for their assistance and support. First, I thank Fred Gifford for his significant time, effort, and support in advising me through this project. I also thank the other members of my Advising Committee: Joe Hanna, Judy Andre, and Len Fleck. Their suggestions significantly improved this work. I also thank the remaining faculty and staff at MSU’s Center for Ethics and Humanities in the Life Sciences: Tom Tomlinson, Libby Bogdan—Lovis, Howard Brody, Clayton Thomason, Jan Holmes, and Laurie Rashid. Their commitment to bioethics and to teaching has influenced me greatly. As scholars, they are committed to improving the world and to taking risks. As people, they are wonderfully supportive. In both ways, they have taught me greatly. My graduate school experience was richer because of my colleagues. Allison Wolf, Jennifer Benson, Crista Lebens, Sonya Charles, and Tricha Shivas were friends, teachers, and supportive critics. Numerous other MSU faculty and students, more than can be named here, deserve my thanks. Joe Totherow graciously discussed numerous parts of this dissertation with me. Shelly Cote, a constant friend, supported my decision to leave the science lab behind while she remained. My parents, June and Warren DeCoster, deserve special thanks. Their encouragement allowed me to pursue a wide range of academic paths, even if they did not always understand why or where it would lead. Finally, Michael Nothnagel provided more assistance, patience, and support than I can fully articulate. To all these people, I am thoroughly grateful. Those named above and many more helped me through this project; all errors, of course, are my own. iv LIST OF FIG CHAPTER 1: EXPLLVATI Case 3: J Illlmductior Medical E x 1 Medical Ex; Explanatory Metaphor A LS€ful . 1~EXp TABLE OF CONTENTS LIST OF FIGURES ......................................................................................................... viii CHAPTER 1: WHY WE NEED AN IMPROVED THEORY OF MEDICAL EXPLANATION S .............................................................................................................. 1 Cases ............................................................................................................................... 1 Case 1: Julia’s Ulcer ................................................................................................... 1 Case 2: Uncle Bill’s Stroke ......................................................................................... 2 Case 3: LS. and “Unexplained Medical Disorders” ................................................... 3 Introduction ..................................................................................................................... 3 Medical Explanations and Bioethics ............................................................................... 7 Medical Explanations and the Philosophies of Science and Medicine ......................... 10 Explanatory Goals and Metaphors ................................................................................ 15 Metaphors as Explanatory ......................................................................................... 16 A Useful Metaphor about Explanations: Explanations as ‘Planing’ Activities ....... 18 l—Explanations should improve understanding .................................................. 18 2—Explaining is a constructive activity ............................................................... l9 3—Explanations have goals or intended uses ....................................................... 19 Limits of ‘planing’ metaphor .................................................................................... 21 Chapter Summaries ....................................................................................................... 22 CHAPTER 2: A BRIEF OVERVIEW OF SCIENTIFIC EXPLANATION THEORIES ........................................................................................................................................... 26 Introduction ................................................................................................................... 26 Hempel’s DN Explanations ...................................................................................... 27 Kitcher’s Unifying Explanation ................................................................................ 30 Wesley Salmon’s Mechanistic Explanations ............................................................ 32 Van Fraassen’s Pragmatic Explanations ................................................................... 35 Schaffner’s Middle Range Explanations .................................................................. 36 Railton’s Ideal and Non-ideal Explanations ............................................................. 38 Conclusion .................................................................................................................... 44 CHAPTER 3: A CRITIQUE OF THAGARD’S THEORY OF MEDICAL EXPLANATIONS AS CAUSAL NETWORK INSTANTIATION ................................ 45 Introduction ................................................................................................................... 45 Medical Explanations as Clinical and Research Explanations ..................................... 47 Reevaluating the Clinical/Research Distinction ........................................................... 50 Research Explanations .............................................................................................. 51 Clinical Explanations ................................................................................................ 52 Thagard’s Rudimentary Explanation Schema .............................................................. 54 Hierarchical Organization of Disease Explanations ................................................. 57 Medical E.‘ Clinical Benefits :1: Weaknesse DCfiIllIlC CNl 8: [J CNFSF; Explanat Navig Navig. The Subt To“ ant St Designin Determininl Conclusion CHWER 4; AVD MEDIC, IntTOdUCtiOr Gender and Gender, E Gender R Gender R “OUSideriI lIllCian5 Patients 35 C Clinjcj allem i PJIiEm ; aliem ; The Clix The First ; e sewn Medical Explanations as Causal Network Instantiations .............................................. 60 Clinical Usefulness of CNI ....................................................................................... 65 Benefits and Strengths of CNI Approach to Medical Explanations ............................. 66 Weaknesses and Problems with CN I ............................................................................ 68 Definition of Environment ........................................................................................ 69 CNI & the Myth of Ideal Explanations ..................................................................... 75 CNI’s Failure to Consider Seriously the Difficulties of Generating Clinical Explanations .............................................................................................................. 77 Navigating Levels and Details .............................................................................. 78 Navigating Pragmatic and Contextual Considerations ......................................... 78 The Subtractive Method for Clinical Explanations .................................................. 80 Towards Strategies for Correcting Clinical Explanations ............................................ 82 Designing Explanations & Keeping Explanatory Space Open ................................. 82 Determining the Epistemic Agents of Medical Explanations ....................................... 86 Conclusion .................................................................................................................... 88 CHAPTER 4: WHO EXPLAINS DISEASE?: EPISTEMOLOGICAL COMMUNITIES AND MEDICAL EXPLANATIONS ............................................................................... 91 Introduction ................................................................................................................... 91 Gender and Explanations .............................................................................................. 95 Gender, Explanations, and Clinicians ....................................................................... 96 Gender Roles, Explanations, and Patients ................................................................ 98 Gender Roles, Explanations, and Topics Warranting Explanations ......................... 98 Reconsidering the Agents of Explanation ................................................................... 103 Communities as Knowers ....................................................................................... 104 Communities Involved in Medical Explanations ........................................................ 111 Clinicians as an epistemic community .................................................................... 112 Patients as an epistemic community ....................................................................... 113 The clinician-patient team as epistemic community ............................................... 117 Patient as inquisitor: ............................................................................................ 117 Patient as providing contextual clues: ................................................................. 118 Patient as providing personal narratives: ............................................................ 119 The Clinician-Patient Team as a Community of Knowers: ................................ 120 The First Problem ................................................................................................... 124 The Second Problem ............................................................................................... 128 Conclusion .................................................................................................................. 129 CHAPTER 5: PRAGMATICS OF MEDICAL EXPLANATIONS ............................... 131 Introduction ................................................................................................................. 131 Sorting Through Questions: Distinguishing Between Interrogative Statements and Why-Questions ........................................................................................................... 135 Van Fraassen’s Pragmatics of Explanation ................................................................. 138 Topic ........................................................................................................................... 141 Contrast Class ............................................................................................................. 141 To Clarify the Question (and the Topic) ................................................................. 144 Allows for Rejection of Improper Questions .......................................................... 147 vi Relevance 1 Backgrount Applicatior Racist bc About About About HIV and The C axe Conclusion CHWR 6: VALUES. l.\" lntroductior The Role of H0“ are Int Personal ' Constitut Kuhn‘s 'Sta Longino‘s \ HOW DOES I. Explanation Accurate}- anfulnc EVernal ( SimpllCitx onClUSlon . CHWER 7; BIBLIOGRM; Relevance Relation ..................................................................................................... 150 Background Knowledge .............................................................................................. 159 Applications in Clinical Practice ................................................................................ 164 Racist boss and Uncle Bill: The use of Social Determinants of Health in R ...... 164 About the rejection of questions: ........................................................................ 166 About the Clarification of questions: .................................................................. 167 About Social Determinants of Health as Explanatory: ....................................... 167 HIV and “Why am I sick?” ..................................................................................... 168 The Case of 1.8. Revisited ...................................................................................... 169 Conclusion .................................................................................................................. 180 CHAPTER 6: .................................................................................................................. 182 VALUES, INTERESTS, AND MEDICAL EXPLANATIONS .................................... 182 Introduction ................................................................................................................. 182 The Role of Interests in Explanations ......................................................................... 184 How are Interests and Values Important to Medical Explanations? ........................... 187 Personal Values ....................................................................................................... 189 Constitutive Values ................................................................................................. 190 Kuhn’s ‘Standard’ Scientific Values .......................................................................... 192 Longino’s Values of Feminist Science ....................................................................... 195 How Does the Articulation of Values Improve Evaluation and Creation of Medical Explanations? .............................................................................................................. 204 Accuracy & Empirical Adequacy ........................................................................... 205 Fruitfulness versus Diffusion of Power and Applicability to Human Needs .......... 208 External Consistency versus Novelty ..................................................................... 215 Simplicity versus Ontological Heterogeneity ......................................................... 219 Conclusion .................................................................................................................. 223 CHAPTER 7: CONCLUDING REMARKS ................................................................. 224 BIBLIOGRAPHY ........................................................................................................... 228 vii LIST OF FIGURES Figure 1 - Dr. Taylor’s Diagram of the Possible Causes of Julia’s Ulcer ......................... 1 Figure 2 - Thagard’s General CNI for Duodenal Ulcers ................................................. 64 viii Ml) Cases Consider the it filoesses. Julia. a 91- Taylor imn dUOdenal ulcer Taylor’s lzm c: that ShOWS [he . Anhnlls Of or pamflll COHdeI ”63“)" USe of leg“ EBPirin) Figure ' CHAPTER 1: WHY WE NEED AN IMPROVED THEORY OF MEDICAL EXPLANATIONS Cases Consider the following cases in which different people try to understand their respective illnesses. Case 1: Julia’s Ulcer Julia, a lawyer, presents with severe, continued heartburn and a loss of appetite. Dr. Taylor immediately recognizes that Julia is presenting the classic symptoms of a duodenal ulcer. Julia asks the likely question, “Well, what caused it?” Since this is Dr. Taylor’s last case of the day, he takes a few minutes to sketch out the following diagram that shows the causes of ulcers: Arthritis or other Genetic predisposition Environmental factors painful condition (e. g., to increase acid (e. g., smoking, stress) secretion, rapid gastric l emptying, infection) Heavy use of NSAIDs l (e.g., aspirin) \ Increased acid secretion, Helicobacter pylori rapid gastric emptying, etc. infection 1 Gasiritis Duodenitis / Duodenal ulcer disease Figure l - Dr. Taylor’s Diagram of the Possible Causes of Julia’s Ulcer He then e figure out but Julia \1 Em and red and folks bad. I Niggers to L' black man, \ yellow-red :1 Indian. lain‘ L'ncle leina. beg lacquft’ing‘ an Educated. [me be?“ related to Uncle [ [he funillllre fa He then explains, “Most ulcers are caused by at least one of these tracks; the trick is to figure out which is causing the ulcer for you...” The doctor begins to discuss his diagram, but Julia seems to glaze over and lose interest.l Case 2: Uncle Bill’s Stroke Uncle Boy: “Well it’s like this. There’s white folks. And there’s black and brown and red and yellow folks. Now white folks done treated all us different kinds of colored folks bad. Them doctors did not care enough about me to listen and get it right. We all N iggers to the white folks. So it don’t matter to them doctors whether I’m a Indian or a black man. We all into drugs and alcohol. And because my skin color is a little bit yellow-red and my hair is straight like a Indian’s, they just decided I must be a alcoholic Indian. I ain’t neither. I am a sick black man whose job made his health ba .” Uncle Boy and his brother, Uncle Bill, both worked in a fumiture factory in North Carolina, beginning at the age of 13. Over the years they finished fumiture by sanding, lacquering, and spraying, using toluene and other toxic substances. Though not formally educated, Uncle Bill is not naive. He asked his doctors if somehow his stroke might have been related to his many years of exposure to chemicals. Uncle Bill: “The doctors told me right to my face that all them years working in the furniture factory couldn’t have done my health no good, but none of the doctors was willing to put anything on paper. If the doctors work for the company, then the company got the doctors in their backpocket. And even if they don’t work for the company, then most of them is too scared to say anything ‘cause they know that the fumiture owners is the big bosses in town. The factories use up all us poor people, not just black people. Then they throw us out and don’t want to give us no benefits. You talking about access to ' Image taken from and case inspired by Paul Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms,” Minds and Machines 8 (1998): 61-78. Thagard continues this example in How Scientists Explain Disease (Princeton, JN: Princeton University Press, 2000). health care in that book. Yeah we got a little bit of access, but we got to fight like hell for it, ‘cause the company controls it and tries to give us as little as they can get away with.”2 Case 3: J.S. and “Unexplained Medical Disorders” J .S. is a white, 55-year-old woman. She is married and has three grown children. Her husband is kind but mostly interested in his friends and his work. Mrs. S has always worked hard, the last 15 years in a hospital laundry. Her mother-in-law has Alzheimer’s disease and needs increasing care every month. Over the last 10 years, J .S. has suffered from muscular pain, fatigue, depression, and constipation. She has seen her primary care physician for countless visits and has been through several tests, referrals, and therapeutic interventions, all of them without benefit. Her medical record displays diagnoses like “chronic pain syndrome,” “irritable bowel,” and “somatization disorder?” The physician is exhausted and annoyed and feels uncomfortable when J.S. is on his schedule. J.S. feels disappointed and humiliated, yet she hopes that the physician will ultimately be able to provide some relief for her.3 Introduction I take the three cases above to be examples of requests for medical explanations that have gone awry. Here, medical explanations are conversations between doctors and patients about patients’ health and disease. Standard theories of medical explanations, drawing from philosophy of science literature on scientific explanations, fail to accommodate these cases: some theories fail to identify these as problems involving explanations, while others of these theories fail to provide remedies. 2 Annette Dula and Sara Goering (eds), “It Just Ain ’1 Fair”: The Ethics of Health Carefor African Americans (Westport, CT: Praeger Paperback, 1994) 2-3. 3 This case is taken from Kirsti Malterud. MD, PhD, “Symptoms as a Source of Medical Knowledge: Understanding Medically Unexplained Disorders in Women” Family Medicine 32 (2000): 603-1 I. See also Kirsti Malterud, Lucy Candib, and Lorraine Code, “Responsible and Responsive Knowing in Medical Diagnosis: The Medical Gaze Revisited” NORA: Nordic Journal of Women ’s Studies 12. l (2004): 8-19. patients ‘1 cases. th.’ expianati remedies. communit hostile clir failure due Explanation Problematic l dis. examples ul mlSUIldCrslm ePistemic age HE general .4 dimming ill Case Of L'nde Generally, doctor-patient conversations about health and disease go well, and patients gain important information for making further healthcare decisions. In the above cases, that communication process has failed, and my arguments regarding medical explanations are meant to identify problems in the explanatory processes and provide remedies. Some would argue that the problems in these cases are the result of failed communication practices, such as giving too many details in the case of Julia’s ulcer, or a hostile clinician unwilling to listen to his patient in Uncle Bill’s case, or explanatory failure due to limitations of biomedical sciences in J .S.’s case. Some may argue the explanation itself is not the problem; it is the delivery of the explanation that is problematic. I disagree with such interpretations of these cases. I will argue that these cases are examples where the underlying structure of such medical explanations has been misunderstood. In the case of Julia’s ulcer, the clinican is wrongly assumed to be the only epistemic agent—that is, he wrongly assumes only he can generate such an explanation. His general strategy of explaining the detailed causal process for all ulcers, and then eliminating the threads that are not relevant to Julia’s case, is also problematic. In the case of Uncle Bill, there are again questions about the power dynamics involved in determining who can generate medical explanations. In addition, this case raises the issue that we often lack ways to discuss social forces like racism as explanatory of illness. In the case of J .S., we have a genuine question as to what would count as a good explanation. The causal factors that have led to J .S. feeling poorly are still under debate. The doctor has been asked to explain why J.S. is ill, and presumably he must respond somehow. Yet because he is uncertain how to respond, he feels frustrated. J .S. herself feels frustrated, having seen numerous clinicians and received multiple, often conflicting, explanations of her illness. A good theory of medical explanation will help sort out some of the misunderstandings between I .S. and her doctor. problen' epistemi relex‘ant informati goal of th Th constitutes medical ex Second. wk some Of the explanation like? In reg resP'OUSes to l Queer?! “hi problems-o). at heath and disc Because of the misunderstandings raised by these cases, a number of common problems arise: the topic to be explained may be misidentified; the proper authoritative epistemic subjects capable of contributing to the explanation may be misidentified; relevant causal information may be ignored while too much attention is given to other information; or ensuring clinicians’ understanding may come at the expense of another goal of the explanation—improving patients’ understanding. This project is an attempt to answer four different questions. First, what constitutes a “good” medical explanation? For this, I will focus primarily on the goals of medical explanations, and when are they effective in clinician-patient communication.4 Second, why have standard theories of medical explanation failed to recognize or remedy some of these problems? Third, who are the epistemic agents who can generate medical explanations? And finally, what would an improved theory of medical explanation look like? In regards to the first question, good medical explanations are clinicians’ responses to patients’ why-questions about their health and disease (e.g., Why do I have cancer?, Why has my T-cell count fallen? Why does my diet influence my stomach problems?), and the explanatory responses improve patients’ understanding about their health and disease. In medical explanations, clinicians must properly craft their responses, making judgments about what information the patient is seeking, how much information and detail to provide, how many risks and side effects should be mentioned. A number of books have discussed clinician-patient communication strategies.5 Regarding the second question, standard theories of medical explanations have begun with a focus on scientific and research goals, rather than on patient understanding. 4 By “good”, I mean medical explanations that are part of clinician-patient conversations about health and are effective in improving patients’ understanding about their health, rather than, say, “good” as “empirically accurate” or other possible conceptions. 5 How to Break Bad News by Robert Buckman is a great clinical guide. At a more abstract level, an overview of clinician-patient communications, common difficulties, and moral and legal challenges are addressed in much of bioethics, but are strong themes in books like Jay Katz’s The Silent World of Doctor and Patient, or David J. Rothman's Strangers at the Bedside. Does this r but inapprc patients. Tl "translator" challenge. l Other factor problem of j explanation This explanations researchers (; resEarthen). Eefleratin g an. exPlc‘mzttion h; “ho “'th in c Finall) from [lit Work medical 9pr” and emetic re. Rimming to t Does this mean such explanatory strategies are wrong? I will argue they may be correct, but inappropriate for developing medical explanations shared between clinicians and patients. These “standard” medical explanations put clinicians in difficult position of “translator” of information. Even if a benevolent clinician is assumed, it is a difficult challenge. Patients will vary on their needs, their ability to understand, and a number of other factors. But standard medical theories have not worried about this, seeing this as a problem of patients’ failure to properly receive explanation, rather than a problem w/ how explanation is generated. This raises the third question: who are the agents involved in generating medical explanations? Standard accounts have focused primarily upon the role of biomedical researchers (as scientists), and only secondarily upon clinicians (as stand-ins for researchers). I will argue this misidentifies the active epistemic agents who participate in generating and evaluating medical explanations.6 Primarily, standard theories of medical explanation have ignored the role of patients as active, epistemically authoritative agents who work in conjunction with clinicians. Finally, I will argue for an improved theory of medical explanations. Drawing from the work of Bas van Fraassen and feminist philosophers of science, I will argue medical explanations are best understood as pragmatic explanations. They are contextual and erotetic responses, thus retaining scientific rigor and clinical expertise while also responding to the explanatory needs of patients. 6 Note that in my work, I do not mean for the phrase “generate and evaluate medical explanations” to be synonymous with philosophy of science accounts on discovery of explanations or of justification of explanations. Instead, I mean the creative process by which clinicians and patients develop explanations through their dialogue, often in that face-to-face encounter within a clinical setting. I will not explore at great length whether there is a logic to discovery or whether this occurs by chance, as some philosophers of science have inquired about other projects. My work reflects a different set of concerns. Med. literatur discussi of heal-ti then pla‘ Vt clinical b; oommunit her doctor bl a Viral i remmmem she has not Would be in Went Ffima be Could her} example. UK PfiIiem'S lnab explanatio“ tl Mica] errpt» regarding he;l Medical Explanations and Bioethics The topic of medical explanations has had nearly no discussion within bioethics literature, and I argue that this oversight requires correction. For the purposes of this discussion, I take bioethics to be the philosophical and moral analysis of clinical practice, of health care policy and law, and biomedical research. I argue that medical explanations, then, play a crucial role in first and third of these themes.7 Why is a discussion of medical explanations and explanation theory relevant to clinical bioethics? Medical explanations are a central aspect of doctor-patient communication.8 Consider the common misunderstanding that arises when a patient visits her doctor because of a common cold. The doctor explains that the cold is likely caused by a viral infection, and that she should start to feel better after another week. He recommends rest and fluids. The patient, though, is unsatisfied; she wants to know why she has not been prescribed antibiotics to fix her cold. The doctor explains that antibiotics would be ineffective in this case. Often, such a response will work. But suppose the patient remains unhappy. She feels her doctor did not do everything he ought to and that he could help her feel better if he wanted to. So, there is a failure of explanations in this example. Locating the source of this failure may involve both parties, perhaps the patient’s inability to receive the explanation, and the doctor’s inability to generate an explanation that is both meaningful and accurate for this patient. An improved theory of medical explanation may benefit general conversations between clinicians and patients regarding health and disease. Consider also that the moral importance of medical information has been a major topic of concern for bioethics over the last few decades. The importance of “informed 7 Bioethics is importantly interdisciplinary by its very nature. My goal here, though, is to focus on philosophical approaches to bioethics, drawing from both moral inquiries and philosophy of medicine and philosophy of science. L. M. L. Ong et al, “Doctor-Patient Communication: A Review of the Literature,” Social Science and Medicine 40 (1995): 903-9l8. consent" as conversatior infonnation effects of ti: aspect to sup medicine. In discussion 01 information : failed to take interests in m The n “Pm-k sense a; informed con elllpltatsis on t bl a belief the “my OUICOm of [he inform Clinician dCCit lOllowed d0t‘t marine er; I P3361113? Yet re 1 clinicianS “'lij Lapel). [liege titre left Una ‘\ .lajt K2112, 7715 consent” as one of the primary principles of bioethics seems to be ubiquitous in conversations about ethical patient care. In brief, the idea is that patients have a right to information about their health and the foreseeable consequences of treatment (e. g., side effects of drugs, negative consequences of surgery). This information is an important aspect to supporting patient autonomy and preventing paternalistic decision making in medicine. In this section I want to highlight some of the key themes connecting the discussion of explanation theory in philosophy of science to bioethical considerations of information and understanding by both clinicians and patients in medicine. Bioethics has failed to take up in a strong way analysis of explanation theory despite the shared interests in understanding and understanding well. The moral consideration regarding patients’ grasp of their health can be read in a weak sense and in a stronger sense. In the weaker sense, the moral worries about informed consent have largely settled as legal concerns. Thus, there has been significant emphasis on topics of disclosure and consent. The idea of informed consent is motivated by a belief that patients have a right to information about their health status, risks, and likely outcomes in order to make their own, well-reasoned health care decisions. The idea of the ‘informed’ patient is a rather recent historical development. Historically, the clinician decided what information to provide to the patient. A good patient accepted and followed doctors’ advice without question. As medical science improved over the century and a half, clinicians faced a new question: whether and when to share information with patients.9 Yet recent historical legal developments have concluded that in some cases, clinicians withheld information important to patients’ autonomous decision making. Largely these were clinical trials (rather than doctor-patient engagements) where patients were left unaware of the dangers they faced. Out of such abuses, patients generally 9 Jay Katz, The Silent World of Doctor and Patient (New York: Free Press, 1984) , esp. Chapters l and 2. developer and abilitj Through [1 justice and But moral treat. lIll-(Drilled p; “11105 hay accelmtnce. (Of healihCQ developed a distrust of physicians. The disagreement is largely about who has the power and ability to know. As one author summarized Katz’s overall themes: Patients can’t trust their physicians to act in their interests, while physicians, trained to believe in what amounts to a self-fulfilling prophecy—that patients are incapable of making informed, intelligent, rational decisions about health care— can’t trust patients to act in their own interests. The result is that patients are essentially disenfranchised. Stripped of power and control in medical decision- making, their interests and values are ignored when they matter most: in matters of life, death and well-being.” Through the process of seeking consent, some clinicians undermine patients by ignoring justice and autonomy concerns. But there is a richer sense in how the role of information has been a concern for moral treatment of patients, one that has less of a legalistic turn for the basis of an informed patient. While Katz locates a change in the doctor-patient relationship, other authors have located the change differently. As patients shed the role of passive acceptance of clinical decisions, they developed a new role, that of an informed consumer (of healthcare), one that is more informed and more savvy. Physicians often encountered a new type of patient, particularly among young and better educated men, and even more frequently, women, who had assimilated a new message: rather than comply dutifully with your doctor’s orders, be “alert to your responsibility in the relationship, just as you would in any other adult relationship where you are purchasing services.”1 There is a new consumerism in medicine, perhaps in addition to a different legal relationship between clinician and patient. This has caused patients to seek more individualized care that meets their needs (rather than the needs of a hypothetical “average patient”). Thus, the role of clinicians and patients has changed considerably with regard to how information is generated and shared. Clinicians can be viewed in the '0 L. Syd M Johnson, “Review,” http://mentalhelp.net/poc/view_doc.php?id=2066&type=book&cn=72. accessed September 9, 2005. it David J. Rothman, Strangers at the Bedside: A History of How Low and Bioethics Transformed Medical Decision Making (New York: Basic Books, 1991), 2. role of 3 Patients consume with reg: changed. the past. c why such complex. DC: explanatior inform pati listing ”.ka agll'fn path will arqu C be Critical of [0 C0ndemn. Patient mm.e and remedies role of a “hired expert”, rather than the authoritative and paternalistic role of “priest”. Patients are emerging—both empowered and with greater knowledge—as critical consumers of goods and services. This has changed the roles of both parties, especially with regard to authority. This also means communication practices have generally changed. Rather than the clinician providing orders for a healthy lifestyle as they have in the past, clinicians must now find means to explain to patients their recommendations and why such advice is best heeded. Clinical communication in general has become more complex. Despite the emphasis placed upon the moral importance of clinicians providing explanations to patients, there is arguably a lack of guidance for how physicians ought to inform patients in this “richer” sense. General communication will require more than listing risks and benefits of a given treatment. How do clinicians know what information a given patient needs or wants? What level of detail? These practical considerations, I will argue, can benefit from a clearer sense of medical explanation. While at times I may be critical of physicians and how they approach interactions with patients, my goal is not to condemn. Instead, I want to identify the numerous complexities involved in doctor- patient conversations about health, the many pitfalls that may impair such conversations, and remedies available through an improved understanding of medical explanations. Medical Explanations and the Philosophies of Science and Medicine Although philosophy of science has developed a significant body of research on scientific explanations, work on medical explanations has rarely taken center stage. Because of this, I will argue that situating discussions about explanations within the scientific domain of medicine rather than one of the other sciences provides interesting new challenges and insights. I identify four themes below, themes that I will continue work in later chapters. 10 bier: gene: also b science accomrr slipping S. Ciplanati't Peers: scie allplopnate lbt‘ context dl-‘COUrse. h (Lied bV non. linens will Smitilllil‘lc figf Sttmtific la»- 1 “Iranian. First, medical explanations raise important questions about the relationship of biomedical science to clinical practice. For example, should explanations of ulcers be generated to explain all persons who develop ulcers, or can individualized explanations also be generated? Here understanding doctor-patient explanations will raise a number of questions about ideal explanations and contextualized explanations. Some may argue that explaining the diseased state of individuals is part of the “art” of medicine, rather than the science of medicine.12 I will argue clinical explanations can be contextualized to accommodate the needs of individual patients, remain scientifically rigorous, and avoid slipping into relativistic explanations. Second, medical explanations raise important questions about the agents of explanations. Primarily scientific explanations have been described as shared between peers: scientists generate explanations to share with other scientists. Although this seems appropriate to standard scientific explanatory projects, new concerns arise when changing the context to medical explanations. Medical explanations are often part of such scientific discourse. Interestingly, though, medical explanations between doctors and patients are used by non-peers: physicians may have significant scientific background, but many patients will not. So, a problem arises that medical explanations must have a level of scientific rigor, but must be utilized by non-peers (those scientifically trained and scientific lay persons). A possible problem arises when the clinician is faced with the task of “translating” the explanation from a scientific context to a lay context.'3 I will argue, though, such translation strategies are flawed and unnecessary. Medical explanations can be generated in a shared context between clinician and patient. Thus, the translation is generally unnecessary. '2 Samuel H. Greenblatt, “Limits of Knowledge and Knowledge of Limits: An Essay of Clinical Judgment” The Journal of Medicine and Philosophy 5. l (1980): 22-29. This issue is devoted to the theme of “Understanding and Explanation in Medicine.” '3 Again, some may consider this process of translation to be part of the art of medicine. While I believe that clinicians and patients can learn to better communicate about health, I will argue that the “translation” approach is flawed in a number of fundamental ways, and thus clinicians should be wary of this strategy. 11 This n epistemic age of medical ex. tum utilize th: here is to oxer Third. be considered involve a land relevant Will t What appeal-5 are daunting l Scient additional ten “Plantation, 5 experiments, I explal'lalions ] causal path“ ; phiSlClan hax Yet 0; POSSibilitiES Unlinked E .. xi ”9131310“, r- mani cl lnici; This move adds another complication: what role do patients play as authoritative epistemic agents in generating and evaluating medical explanations? Traditional theories of medical explanations have viewed patients as sources of knowledge, and clinicians in turn utilize this information in generating medical explanations. But I argue that to stop here is to overlook how patients structure and evaluate medical explanations. Third, medical explanations oblige us to ask what types of causal factors should be considered as explanatory of health and disease. Medical explanations will often involve a tangled knot of information. In some cases determining which threads are relevant will be difficult, as in the case of Julia’s ulcer; in other cases, sorting through what appears to be conflicting information may be required, as in the case of J .S. Both are daunting yet common aspects of clinical medicine. Scientific explanations are often quite complex. Medical explanations have additional tensions, though, such as limited time and data from which to generate an explanation. Scientists may have years to gather better data through continued experiments, but this is not a luxury clinicians have when seeing patients. Medical explanations may be different from scientific explanations in another way. Scientific explanations are often reductive projects: the explainer often tries to identify all possible causal pathways in order to determine the actual causal pathway at work. Julia’s physician has taken this strategy to explain her ulcer. Yet often medical explanations will be projects meant to open up explanatory possibilities, rather than to reduce explanatory options. J .S. is seeking additional and improved explanations of her illness. In some cases, like that of Uncle Bill, the explanatory role of social forces like racism on his health is a primary concern. While many clinicians may see such social forces as outside their expertise as clinicians, they are nonetheless explanatory in some cases. Thus, medical explanations often have the 12 forces a an impo processe F aslr What patients ( Standard i highly cor exPiiittatio As, CliIIItot be u Cmillet. th 0f SCientitic diMISS in gym acor”litter n hmlliillofp bl 15 This l5 [0 b” “IS I ‘ A ”first 4‘ j goal of “making the invisible visible.”l4 Medical explanations sometimes have the goal of considering new information, calling our awareness to what we had not noticed before, to what we had discounted. While this may make the explanation more complicated, it may also be more honest, more accurate, and more meaningful to patients. One additional source of information that I will focus on is the role of social forces and social determinants of disease as explanatory.15 I will show in Chapter 4 that an important addition to a theory of medical explanations is the consideration of social processes as salient causal factors that are explanatory of individuals’ health and disease. Fourth, medical explanations as a topic of philosophical investigation allow us to ask what level of causal detail is appropriate for explanations shared between doctors and patients (that is, between scientific experts and lay-persons). Here, I disagree with standard theories of medical explanations: I argue that explanations that accurate but highly complex—such that they cannot be understood by patients—are not good medical explanations. As Sandra Harding argues, explanations should not be long, detailed formulas that cannot be understood by people. While not all scientific explanations end up being so complex, the focus on causal chains lends itself to this possibility. Under standard models of scientific explanation—including those of Hempel, Salmon, and Railton, which I discuss in greater detail in the following chapter—a good explanation can end up being quite complex, even to the point that they are incomprehensible to most folks, or require a computer to hold all the relevant information. They see such complexity not as a limitation, but as uncovering the complexity of natural events. If explanations end up quite complex, then better to have complex and ‘complete’, rather than an explanation that is comprehensible but incomplete or otherwise inaccurate. “ This is to borrow a phrase from Rothman in Strangers at the Bedside. '5 This interest in social determinants of disease is not exclusive to the philosophy of medicine. Judith Andre (Bioethics as Practice) discusses how this is also an under-examined (or unexamined) topic in bioethics. See her arguments for the importance of improving discussions regarding poverty and health. 13 Yet this makes the mistake of prioritizing the project of listing complete the causal factors over the importance of understandability. After discussing a rather detailed physics explanation, Sandra Harding lays out the following intuition, which I share, about the nature of explanation: [S]hould we think of a formula so long that only a computer could read it in one hour as an explanation of a type of phenomenon? The answer to this question is “no.” An explanation is a kind of social achievement. A purported explanation that cannot be grasped by a human mind cannot qualify as an explanation. If no human can understand, can hold in the mind, the purported explanation, then explanation has not been achieved. 16 I want to discuss two points about Harding’s comments. First, I find her conception of explanations as “a kind of social achievement” important and thought-provoking. If they are social projects, then who are the participants? What goods do they have in mind when achieving explanations? In what way should we value “completeness” of causal explanations? How do social factors and personal histories influence what we see to be an explanation (and does this change for individuals and for groups)? I return to the themes of these questions throughout the following chapters. The second point raised by Harding’s comments is that explanations, in general, are first and foremost meant to be understood by embodied persons. Although Harding is speaking here specifically about physics explanations, her main point holds for medical explanations as well. Harding emphasizes that an explanation must be understood in order for it to be explanatory. Explanations are not theoretical accomplishments, but are a means of improving our understanding. In medical explanations, in particular, it is not enough that one person (typically a clinician) is able to understand the explanation. There is a different requirement for medical explanations: they must be understandable to most patients for the explanation to have occurred. Such a requirement of understandability is '6 Sandra Harding, The Science Question in Feminism (Ithaca: Cornell University Press, 1986), 45, original emphasis. 14 Mun"... not to minimzl patient’s sakel The difficult p patients is u h l l I exemplify the Projects: thes not to minimize medical explanations, say, by “dumbing down” the explanation for the patient’s sake. Instead, something more difficult is going on with medical explanations. The difficult process of making complex information understandable and usable to patients is what I am interested in. I will argue that a pragmatic theory of medical explanation best accounts for these considerations. Such a theory will show that explanations are generated or produced for and by people; explanations are not unearthed or discovered facts of the universe. Fifth, medical explanations (as a certain type of scientific explanation) better exemplify the point that generating and evaluating explanations are not just epistemic projects; they are moral/political projects as well. Rather than disembodied projects about explaining electron movement or genetic heritability, the context of medical explanations highlights the fact that explanations are generated by and utilized by embodied and social situated persons. Beyond the epistemic considerations for a “good” explanation that others have proposed (such as accuracy, fruitfulness, and breadth of an explanation), medical explanations are directly useful to patients’ health care decisions.17 Thus, the project of generating “good” explanations must also benefit patients. Power dynamics in generating the explanation must be explored—patients ought not be silenced or excluded from medicine, nor should clinical expertise allow clinicians to disempower their patients. For these arguments, I will draw heavily from feminist contributions to the philosophy of science. Explanatory Goals and Metaphors Many authors have argued about the proper structure of explanations, or what information counts as explanatory. '8 For instance, J aegwon Kim argues that an '7 In Chapter 6 I take up these epistemic virtues argued for by Kuhn by articulating their strengths and weaknesses for medical explanations. '8 Two useful texts: first, see Peter Achinstein, The Nature of Explanation (New York: Oxford University Press, 1983). Second, for a useful history of the concept of explanation in which he intentionally discusses 15 explar expla: predic. con-cc; and en prm'id: explanz epistem disease, Harding 1 Before [h kind 0f Wt areatipe explanation ought to relieve epistemic puzzlement. Kim writes, “When we look for an explanation of an event, we are typically in a state of puzzlement, a kind of epistemic predicament. A successful explanation will get us out of this state.”19 In this way, Kim’s conception of the goals of explanation echoes Harding’s earlier notes about the applied and embodied aspect of explanatory goals. Salmon instead argues that explanations provide an understanding of the underlying causes of an event. I will argue that medical explanations that are part of clinician-patient conversations should relieve patients’ epistemic puzzlement, rather than provide understanding of the causal mechanisms of disease. That is, I find as a beginning point for understanding medical explanations Harding’s and Kim’s goals for explanation to be more useful than are Salmon’s. I will return to explication of the structure of an explanation in the next chapter. Before this, I want to take a closer look at the goals of explanations. That is to say, what kind of work should we expect them to do for us? A common theme is that explanations are a type of tool involved in improving our understanding. Metaphors as Explanatory A discussion here on the use of metaphors and their connections to explanations will be useful. First, I will discuss how metaphors are one explanatory strategy to improve understanding that is often used in medicine. Second, I discuss a metaphor meant to help us better understand explanations themselves. Metaphors are common tools of communication, ways of getting others to understand what you mean. Metaphors compare two topics and highlight their similarities. Doctors often use metaphors in conversations with patients as a way of getting patients to understand the more important facets of a conversation. For example, a doctor is trying to explain to his patient that he has an abdominal aortic aneurysm (AAA). The AAA is serious, and warrants surgery in ‘explanation’ both within the philosophy of science and outside that domain, see David-Hillel Ruben, Explaining Explanation (New York: Routledge, 1990). '9 Jaegwon Kim, Supervenience and Mind: Selected Philosophical Essays (Cambridge: Cambridge University Press, 1993) 254. 16 the near I understar it" So. r2 explains 1 your hear lino“; sor appears; (3 “ant to gr Mmawa Certain util its n0t heir hose. how ; ftpair PTOCt “hi has “5 limit Undergrad“ Surger3'10 tl. ademjnal SI the near future. The patient, an auto mechanic by profession, has a rather poor understanding of biology and medicine, and the doctor is worried that he isn’t “getting it.” So, rather than give another detailed description of vascular physiology, the doctor explains the situation with a metaphor: “There’s a large artery that carries blood from your heart to your legs, just as a radiator hose carries fluid through a car’s engine. As you know, sometimes a radiator hose develops a weak spot. When that happens, a bulge appears; over time the bulge may rupture. That’s what’s happening inside of you. We want to go in and replace that section of the hose before it ruptures.” The idea of a hose with a weak spot is common and understood by the patient. Thus, the metaphor has a certain utility in improving the patient’s understanding of his medical condition despite its not being perfectly accurate. The patient is familiar with the ideas of a weak spot in a hose, how it will bulge under pressure, the importance of avoiding a rupture, and the repair process. While good metaphors can be wonderfully illuminating, even the best metaphor has its limits. Metaphors begin to fail at the point where the comparisons fall apart, and understanding is not improved. The above metaphor fails to capture the overall risk of surgery to the patient. Replacing a radiator hose does not endanger the whole engine, yet abdominal surgery poses a number of risks and complications that may arise for the patient during and after surgery. 80, metaphors as an explanatory strategy are helpful in a limited way. They are often used to explain aspects of disease. But the limitation of metaphor as an explanatory strategy prevents them from being the basis for all explanations in medicine. Here, I will not develop a theory of metaphor, nor do I develop a strong theoretical link between explanation and metaphor.20 Instead, I want to point out that the 2° For such work. see, for instance, Susan Sontag, Illness as Metaphor & AIDS and Its Metaphors (New York: Picador, 2001); Richard Boyd, “Metaphore and Theory Change: What is ‘Metaphor’ a Metaphor For?” in Andrew Ortony (ed.), Metaphor and Thought (Cambridge: Cambridge University Press, 1993), 356-408; T. S. Kuhn, “Metaphor in Science” reprinted in Metaphor and Thought. 403-19. 17 use of meta; conversatior metaphor th An ll Which remai comes from “Nurture: li meaning "‘lC blame.» a Ca. ~ 1 fine instead been] Stretching 0t explanaIIOn ‘ 1‘5 KCCp puzzlement.. hot as a Car: Shaping and Explanation 331mm “a 21 numb?! of (j Other-3 ha Ré‘l‘mi \e Emir, "g Dl ’- . 2 "‘G'tartr." ‘ use of metaphor is often a basic explanatory strategy in medicine, especially in conversations between doctors and patients. In the next section, I want to suggest a metaphor that may elucidate certain aspects of explanations. A Useful Metaphor about Explanations: Explanations as ‘Planing’ Activities An important metaphor involved in explanations—one I find particularly helpful, which remains in the background throughout my discussion of medical explanations— comes from the etymology of the word ‘explain’.21 ‘Explain’ is derived from the Latin explanare: literally, to make level (from ex- meaning “out” or “from” and planus meaning “level, flat”). Here, ‘explanation’ shares a common etymological origin with ‘plane’, a carpenter’s tool used to flatten or smooth wood.22 I find this connection interesting not because it tells us what an explanation is, but instead because it illuminates what we want explanations to do. The metaphor is useful in stretching our imagination about the goals and process of explaining. Imagine, then, an explanation as a tool, much like a carpenter’s plane. Below I identify three themes that I will return to in this and later chapters. I—Explanations should improve understanding Keeping with Kim’s earlier comments that explanations should “relieve epistemic puzzlement”, an explanation should level out “bumps” in our knowledge of the world, just as a carpenter’s plane levels out the rough surfaces of a board. Both are practices of shaping and refining of raw materials, either on wood or on biomedical understanding. Explanations are “planing” activities in our daily lives, which are meant to reduce or smooth away the bumps that otherwise impede our understanding. The bumps can be a number of different things, including a basic ignorance of certain facts or theories. Yet if 2‘ Others have noted the origins of this word in their work on explanations. See Mary Ann Cutter, Reframing Disease Contextually (Netherlands: Klewer, 2003) and Peter Achinstein, The Nature of Explanation. 22 Oxford English Dictionary Online. http://dictionary.oed.com/, accessed September 9, 2005. 18 this were I explanatio Not all inf providing 1 They help the overall 2_ Bot natural “or also an den learns to us« this point. I clinicians a, and they ger [Opic at leng 3x5 As if light lObf‘ B this were the only hindrance, we would need further information, not a theory of explanation. Explanations have a structural component, which I will examine in detail. Not all information will be explanatory, and there will be better and worse ways of providing explanatory information. In this way, explanations are simplifying activities. They help us sort through complex and difficult information, and help us make sense of the overall picture without becoming entangled in minutia. Z—Explaining is a constructive activity Both tools and explanations are human creations, not discovered features of the natural world. Carpentry, as an activity, generates a flat board. Similarly, explanation is also an activity involving agents of explanation. Just as the carpenter is the person who learns to use the plane, people generate and utilize explanations in various domains. At this point, I will assume the basic agents involved in generating medical explanations are clinicians and patients.23 These are the persons who ask explanation-seeking questions, and they generate and evaluate the explanatory answers. In Chapter 3, I examine this topic at length. 3—Explanations have goals or intended uses As the son of a carpenter, I grew up hearing the mantra, “Use the right tool for the right job.” Both tools and explanations have an intended purposes and proper uses. Proper use of tools and of explanations often leads to the success of goals. One question I will return to repeatedly is what does it mean to explain well in the context of medicine. Not all information, even if correct, is explanatory. Some explanations relieve epistemic puzzlement better than do others. There may be multiple explanations that we can use, and we need to know how to choose wisely between them. When discussing goals and intended uses, it should be noted that both tools and explanations can be misused. Some tools are more efficient for a particular purpose than 2’ I will complicate this view of the agents of medical explanations in Chapter 4. 19 are other flathead 3 explanatii of the exp views the hammer ll how we a; due with v most rclev chant to. flplane use. POOrIy. A p epistemic p explanation how effCCtir For: cancel): “h The pallem ‘ WillCh Such Sim books and \ Somwam are others. It is generally better to use a Phillips screwdriver on a Phillips-head screw. A flathead screwdriver may work, but it is often more difficult. In some case, an explanation may work, but require more effort (on the part of the explainer or the receiver of the explanation) to comprehend. At other times, the tool at hand determines how one views the task at hand. Consider the adage, “When you walk around long enough with a hammer in your hand, everything starts to look like a nail.” The tool at hand can shape how we approach some projects. Rather than look for different tools, we may try to make due with what we have available, or what we are comfortable with. These worries will be most relevant to theories of explanation, to be discussed below, that describe explanations as fixed rather than contextual. When misused, both the explanations and tools can cause harm. In woodworking, a plane used poorly can damage the wood. In medicine, harm may come from explaining poorly. A poorly considered explanation may confuse a patient, rather than relieve her epistemic puzzlement. Thus, there is a sense of contextuality to both tools and to explanations. Their ‘goodness’ or ‘badness’ as tools is judged on the job to be done, and how effectively these tools bring about that end. For example, discussions of cancer often invoke the metaphor of “a war on cancer”, which is meant to be helpful in describing our intent to eradicate the disease. This explanatory metaphor, though, has too often resulted in “victims of friendly fire.” The patient’s well-being and quality of life are too often erased from the conversation, for which such metaphors rightly have been criticized.24 Similarly, Emily Martin has argued that the metaphors used by biological text books and scientific literature to describe human gametes (sperm and eggs) typically do so in ways that replicate gender stereotypes of human men and women. Eggs are 2‘ See also Emily Martin’s description of the body as an independent state which continually tries to fight off external attacks, “Toward an Anthropology of Immunology: The Body as Nation State”, in Mario Biagioli, The Science Studies Reader (New York: Routledge, 1999) 358-371. 20 more at distIEs's The Ian} informal ability 0: fertilizat: C been deri Cxplanatit‘ hah'e afgui typically described as passive and waiting for the arrival of sperm, which are described as more active, complex, and goal-oriented.25 The metaphors used—often of a damsel in distress awaiting a knight to save her—limit how we explain the process of fertilization.26 The language chosen to tell this story of fertilization has implications for what pieces of information are seen as relevant, and what is overlooked. The “powerful” swimming ability of the sperm is often overemphasized; the complexity of the egg’s mechanism in fertilization is often underemphasized or ignored.27 Clearly, this is different from the technical definitions of ‘explanation’ that have been derived in the philosophy of science. But even in the non-medical sciences, explanations are tools for understanding. Later in this chapter I will address how some have argued scientific explanations should be structured and evaluated. Limits of ‘planing’ metaphor The metaphor of explanation as ‘a planing activity’, like all metaphors, has limitations. Although I have argued the metaphor is useful in that it points out that an agent—the explainer—must exist, we still know little about who are the agents involved in explanations. Are all agents equal, or is some skill required? Similarly, what is the best material to use in explanations? That is, what stuff is it we use to lessen our epistemic uncertainty? Finally, the idea of a product or endpoint of an explanation is useful. But should we all have the same goals or endpoints? (To continue the ‘planing’ metaphor, do we all seek the same level of flatness?) Some of these problems I will address explicitly. My goal for this etymological analysis and extended metaphor remains this: the beneficial idea of explanations as active, creative processes of ‘flattening’ or ‘smoothing out’ misunderstandings has often been 2’ Emily Martin, “The Egg and the Sperm: How Science has Constructed a Romance Based on Stereotypical Male-Female Role,” Signs, 16 (1991): 485-501. 2" Martin extends her point to include descriptions of both male and female reproductive systems, not just gamete fertilization. This attention (or lack of it) can occur in social-cultural portrayals, biological text book descriptions, or within structures and topics of scientific research and experimental procedures. 21 orerlool formula; medical i‘ explanat problem best cont dissertati strategies examples COHEéCIiO a backgro. be a subse explanatio, (DISCOT‘erj ErPlain D,“ mlSUflderst; Chaliter Paul -. . .5“ “ill >r Jill‘ljn I . ' gflgl}. “muffler“ overlooked. By starting with this understanding of explanation, rather than a structural or formulaic understanding, I believe important questions are raised about the nature of medical explanations that otherwise may be obscured. Not all explanations, though, will be equally helpful in medical practice; some explanatory strategies may plane away our misunderstandings while others generate other problems. I want to explore how philosophical discussion of scientific explanations can best contribute to the generating and evaluating of medical explanations. In this dissertation, space fails to permit a complete description of all the various types and strategies of scientific explanations. I will only point out a few of the more prominent examples28 both for the sake of background knowledge, and for drawing their connections to medical explanations (the primary topic of this project). This will serve as a background to the topic of this dissertation. Medical explanations have been assumed to be a subset of biological explanations, which are themselves a subset of all scientific explanations. These are in large part the explanatory strategies taken up by Ken Schaffner (Discovery and Explanation in Biology and Medicine) and Paul Thagard (How Scientists Explain Disease). I will argue this strategy has misdirected our attention, resulting in misunderstanding the nature of medical explanations. Chapter Summaries Paul Thagard and Kenneth Schaffner have argued medical explanations should be structured as scientific explanations. In Chapter 2, I critique this move as blurring the structural distinctions between two types of medical explanations: biomedical research explanations (from the laboratory) and clinical explanations (developed as part of doctor- patient interactions). Philosophers of science have advocated primarin antic models of disease explanation that focus on complex causal-mechanistic interactions, e.g., between 2” These will include Hempel’s deductive-nomological (DN), Salmon’s mechanistic explanations, Kitcher’s unifying explanations, van Fraassen’s pragmatic explanations, Railton’s ideal and non-ideal explanations. and Schaffner’s biological and medical explanations. 22 the environment, the body, genetics, and infectious agents. Yet ontic approaches remain inadequate as a basis for clinical explanations given that such approaches to medical explanations often fail to meet patients’ explanatory needs. Clinical explanations, in particular, should both reduce patients’ puzzlement about their sickness and inform their health decision-making process. Thagard wrongly argues that his explanatory schema—Causal Network Instantiations, or CNI—is suitable for both research and clinical explanations. Thagard’s proposed schema, though, cannot address a wide array of patients’ explanatory needs. That is, it often cannot answer patients’ why-questions about their health. Finally, this schema disempowers patients by viewing them only as subjects of explanation, rather than acknowledging them as epistemically authoritative subjects capable of participating in generating medical explanations. Towards a response to these limitations, I begin by proposing a preferred understanding of the agents of medical explanation, one that is more complex than has previously been theorized. I argue in Chapter 3 that clinicians and patients generate medical explanations through collaborative efforts. I utilize Lynn Hankinson Nelson’s concept of communities (rather than individuals) as primary epistemic agents. I show how clinicians, patients, and clinician-patient groups shape explanatory requests and responses. This rectifies Thagard’s oversight by recognizing how willing and competent patients often take on active and beneficial roles in generating medical explanations. Rather than rely upon exclusively causal-mechanistic schemas, I argue in Chapter 4 that medical explanations are best understood as erotetic projects (explanations that answer the situated why-questions of the particular patients and clinicians involved). I draw heavily upon Bas van Fraassen’s work on pragmatic explanations in The Scientific Image. This model allows the generation of medical explanations to be collaborative projects that account for both clinicians’ and patients’ interests as shaping medical explanations. Their personal interests shape the why-question asked, as well as the 23 contrast C what inf C. Thus. a Cl the UlllVC‘. Vt neither is propose a at the leve difficultie: level Of ex focusing o; Complexly Seconcl l s Clmficatior explanamr} healthcare d In Cl MOM b explanail-Otis and evaanti This allOWS account for contrast classes and relevance relations (that is, the explanatory concepts that regulate what information appropriately can be considered as part of the medical explanation). Thus, a contextual approach to explanation is superior to meeting such needs, rather than the universal or acontexual approach developed by Thagard and others. While scientific explanations and medical explanations are obviously connected, neither is reducible to the other.29 Drawing from work on scientific explanations, I propose a pragmatic theory of medical explanations. The ultimate goal is twofold. First, at the level of practice, I provide suggestions to clinicians for navigating the real-life difficulties of providing explanations of disease during clinical encounters. Second, at the level of explanation theory, I develop van Fraassen’s theory of pragmatic explanations, focusing on the specific domain of medicine. This raises important challenges. First, the complexly intertwined explanatory roles of clinicians and patients require clarification. Second, I show how the structure of van Fraassen’s explanatory theory allows for clarification of explanatory requests, their occasional rejection of inappropriate explanatory requests, and the development of explanatory responses that inform patients’ healthcare decision-making process. In Chapter 6, I argue for the inclusion of the values of a feminist science (as developed by Helen Longino) as additional pragmatic features that shape medical explanations. These additions illuminate the moral-political components of generating and evaluating medical explanations, while maintaining the explanation’s scientific rigor. This allows guidance for fusing clinicians’ and patients’ interests in asking why- questions about disease. It also broadens the scope of medical explanations to better account for social determinants of health and disease. ’9 See Ronald Munson, “Why Medicine Cannot Be a Science,”Joumal of Medicine and Philosophy, 6 (1981): 183-208. 24 Finally, I return to the three cases described at the beginning of this chapter. I show how an improved theory of medical explanation can better account for the explanatory requests of Julia, of Uncle Bill, and of J.S. 25 A BR lntrod I r approach: been pro; questions some of ti CHAPTER 2: A BRIEF OVERVIEW OF SCIENTIFIC EXPLANATION THEORIES Introduction I provide in this chapter a concise overview of philosophy of science’s significant approaches to explanation theory. I point out some of the significant strategies that have been proposed, and discuss their relative strengths and weaknesses. Some of these same questions will be raised in my discussion on medical explanations, and I will discuss how some of these theories are better suited than others to structure medical explanations. In addition, I will discuss how each author’s choice of scientific domain to examine (whether physics, chemistry, biology. or medicine) influenced the theory’s development. I will also locate my work on medical explanations along this continuum. Here, I would like to make a note on types of scientific explanations, and then identify some helpful distinctions. Generally, scientific explanations are generated as part of various scientific domains, e. g., physics, chemistry, or biology. The work of Hempel and Salmon focused mainly, although not exclusively, on the physical sciences. Kitcher’s work on explanatory unification largely focused on the domain of biology. Interestingly, van Fraassen believed all explanations were structured similarly; the only thing unique about scientific explanations (as opposed to a common, everyday explanation) is that they limit their inquiry and evidence to the domain of science. Explanations involving the life sciences, e.g., biology and medicine, will generally be of greater interest to this project on medical explanations than are explanations involving chemistry and physics. Shaffner’s and Thagard’s works each focus on explanations in the biological and medical sciences. In these domains, there is commonality in the use of theories, say, of genetics, disease, evolution, and function. 26 hum dislll biomr (that i biome distinc in detai categori understa Medical explanations will be primarily relevant to the health and disease of human beings. Yet even in discussing types of medical explanations there are important distinctions to take into consideration. Most importantly, explanations generated in biomedical research can be distinguished from explanations generated in clinical practice (that is, explanations between doctors and patients). Shaffner’s work is primarily about biomedical research; he has virtually nothing to say about clinical explanations. The distinction between research explanations and clinical explanations is one I will develop in detail and show the importance of in Chapter 3. Thagard is interested in both categories of medical explanations, yet his work focuses almost exclusively on understanding biomedical research explanations.3O In my work on medical explanations I will focus primarily upon clinical explanations, yet I will also address research explanations when appropriate. I will draw heavily from the work of Thagard and Schaffner, who I take to be the authors who have best articulated current theories of medical explanation. I will argue that some of these debates regarding the nature of scientific explanations have (wrongly) directed our attention, causing misunderstanding about the nature of medical explanations. Hempel ’s DN Explanations In 1948, Hempel and Oppenheim published the first major work on modern scientific explanations, and for a significant period of time, it was held as the definitive characterization of scientific explanations. Hempel’s deductive-nomological (DN) model of an explanation is a valid, deductive argument. It has two major parts: an explanandum, a sentence “describing the phenomenon to be explained” and an explanans, “the class of those sentences which are adduced to account for the phenomenon.”3 ' The 3° I take up the work of Thagard in detail in Chapter 2. Although he claims his theory of medical explanations can account for both research and clinical explanations, I question its utility, especially in understanding clinical explanations. 3‘ Carl Hempel and Paul Oppenheim, “Studies in the Logic of Explanation, ” in Carl Hempel, Aspects of Scientific Explanation and Other Essays in the Philosophy of Science (New York: Free Press, 1965): 247. 27 CVCIII to 1 covering following E l5 3 dt‘st more reler [hm are re deductive. the CXpian‘ SOUP. the emfillmem. informallm abOut E gh( [0 have ha; my pen ant explanailop Obj eCLS lO'v. event to be explained (E) must call upon general scientific laws—sometimes called ‘covering’ or ‘overarching’ laws—and specific relevant factors. DN explanations take the following structure: L,,L,,L,,...L c,,c,,c,,...c } explanans E } explanandum E is a description of the empirical phenomenon to be explained. L refers to the one or more relevant scientific laws. C refers to the one or more facts or antecedent conditions that are relevant to understanding why E occurred. Because DN explanations are deductive, valid arguments, if the laws (or, nomological premises) and facts cited within the explanans are true, then our derived explanandum must be true as well. Taken as a group, the explanans provides understanding about—that is, it explains by deductive entailment—the explanandum (E). This approach is helpful because it meets a number of our intuitions about what information must be part of an explanation. Presumably, all the information that brought about E should be cited. If the features of the explanans were present, we say that E had to have happened without some additional, intervening factors. For example, if I let go of my pen and it subsequently rolls off my desk, I expect that it will land on the floor. The explanation (or the prediction) of this event would require me to cite that gravity pulls all objects towards the earth (a universal law) and that my pen rolls off my desk at a certain height (an antecedent condition). These factors explain why the pen falls to the floor. The DN model has an additional benefit: the structures of explanations and of predictions are the same. I can explain the falling of my pen (after it has happened) by citing gravity and the fact that I let go of it. I can also accurately predict (before I let go 28 ofthep be SII'Ut allows I may not later. In at some I model in laws in b the laws t biochemi; laws of to Sc IIIOdel. Th refemn g u bi refemn 510nm”. a Shadow (w of the pen) what will happen should I drop it. Both the explanation and the prediction will be structured as a valid, deductive argument under the DN model. Still, we rarely have complete DN explanations in our everyday lives. Hempel allows that there may be ‘elliptical’ explanations and ‘explanatory sketches’ where we may not provide all the relevant laws or provide only generalities that might be filled in later. In these cases, a complete explanation is assumed possible, and the explainer could at some point fill in all the explanatory details or cite relevant scientific laws. Despite the benefits of Hempel’s approach, there are problems that make the DN model inappropriate as a basis for medical explanations.32 First, there are few covering laws in biology and medicine. In biology, evolutionary theory may serve that purpose, or the laws of biochemistry.33 Yet this would require that we derive all biological laws from biochemistry laws. The problem of reduction increases when trying both (a) to identify laws of medicine and (b) to derive them from biochemical laws. Second, a number of counterexarnples have been developed against the DN model. These include the explanation of Mr. Jones’ failure to become pregnant by referring to his use of birth control pill”; Scriven’s explanation of the falling barometer by referring to approaching storms, but the barometer’s falling does not cause the storms35 ; and Bromberger’s explanation of the relationship between a flagpole and shadow (where the length of the shadow is explained by referring to the pole length, but not vice versa).36 ’2 For an interesting discussion of how Hempel’s DN model remains important for philosophers of psychology and biology (those who are non-specialists but interested loosely in causation), see James Woodward, Making Things Happen: A Theory of Causal Explanation (New York: Oxford University Press, 2003) 4. For this group, many believe DN’s rough outline is accurate, despite various problems that still require resolutions. 33 Kenneth F. Schaffner, Discovery and Explanation in Biology and Medicine (Chicago: University of Chicago Press, 1993) 64. 3‘ This example highlights the problem of a valid DN explanation that lacks any reasonable causal mechanism. 35 Again, this example shows the importance of a causal mechanism. Without it, one may wrongly conclude from the evidence that the falling barometer causally explains the storm. 3" This example shows highlights the problem of explanatory asymmetry. 29 explan. what if true the” explan: somehc require use oft Third, adding true but irrelevant information to an explanation may damage its explanatory ability in certain contexts. It is true that “all salt dissolves in water.” But what if we were to say, “all salt blessed by a priest dissolves in water.” While it may be true that all salt (blessed and not blessed) dissolves in water, the blessing is not explanatory. Instead, mention of the blessing may distract by implying that the blessing is somehow relevant to the dissolving of the salt. The hexed salt explanation fills the formal requirements of a DN explanation, but it seems to fail to be a valid explanation given its use of bogus (or incorrect) information.37 Kitcher’s Unifying Explanation Philip Kitcher38 and Michael Friedman39 have advocated that explanations should have unifying ability, which provides understanding. Under explanatory unification, competing explanations are more explanatory to the degree that they minimize the number of unexplained facts or basic premises that scientists are forced to accept in their total picture of nature. Unified scientific explanations increase our understanding by “reducing the total number of independent phenomena that we have to accept as ultimate or given. A world with fewer independent phenomena is, other things being equal, more comprehensive than one with more.”40 For example, both Mendelian phenotypic ratios and the influence of retroviruses (e. g., HIV after cellular infection) can be explained by reference to a common theory—modem genetics. Unification accounts of explanations are sometimes referred to as “top-down” approaches: under such models, explanation begins with our most general scientific understandings, and “flows down” to more specific phenomena to be explained. In this 37 Wesley C. Salmon, Four Decades of Scientific Explanation (Minneapolis: University of Minnesota Press, 1989) 50. 38 See Philip Kitcher, “Explanatory Unification.” Philosophy of Science 48 (1981): 507-53 l; “Explanatory Unification and the Causal Structure of the World,” in Philip Kitcher and Wesley C. Salmon, Scientific Explanation. Minneapolis: University of Minnesota Press, 1989: 410-505; and The Advancement of Science (New York: Oxford University Press, 1993). 39 Michael Friedman, “Explanation and Scientific Understanding,” Journal of Philosophy 71 (1974): 5- l9. 4° Friedman, “Explanation and Scientific Understanding,” 15. 30 utilize 5 PCS.» CIITTTC Ur unificat would h equahyz hiomedic 0i CXpIan BUI not a} “mild do Unificatin damn way, there are similarities between unifying and DN explanatory theories. But Hempel’s DN model does not place a greater value upon the use of fewer scientific laws, in contrast to Kitcher’s unifying model. Two potential problems make the use of explanatory unification unhelpful as a basis for medical explanations, but for two rather different reasons. First, biomedical science is by its nature rather heterogeneous. It pulls from a vast set of scientific theories. This complexity makes the project of unification in medicine significantly more complex. Not only might we need to first unify the various sciences—physics, chemistry, biology and genetics—we would then need a unification theory of these threads as they are utilized by biomedical science. This is only to say that if a unifying theory of explanation is possible, it may also require other prior unifying projects, thus making it even more a difficult project. The second is more directly important to the applicability of explanatory unification to medical explanations. Even if such a unification project were possible, it would have a limited scope, and not all medical explanations would likely benefit equally. A unifying theory of explanation in medicine would likely benefit the biomedical scientist. Unification is a feature of theoretical understanding about patterns of explanation, and how these patterns come together to provide scientific understanding. But not all explanations occur as part of lab research. A unifying theory of explanation would do little to help the explanations that occur between doctors and patients. Unification, which is about patterns of explanation, has direct benefit to the explanation of an individual case. If, as I am arguing, we need a theory of explanation that can benefit doctor-patient conversations about disease, explanation as unification seems to hold little promise. 31 Here S Upon tl Semanti kDOWIfi memod . Wesley Salmon ’s Mechanistic Explanations Wesley Salmon begins his alternative for an explanatory theory by arguing against the logical structure of Hempel’s DN explanation. It is my view that the attempting to give a logical characterization of scientific explanation is a futile venture, and that little of significance can be said about scientific explanation in purely syntactical or semantical terms. I believe, rather, that what constitutes adequate explanation depends crucially upon the mechanisms that operate in our world. In all of this there is—obviously—-no logical necessity whatsoever.41 Here Salmon explicitly moves away from theories of scientific explanation (a) that rely upon the logical necessity emphasized in Hempel’s DN explanations, and (b) that involve semantic or natural-language theories of explanation. Salmon instead emphasizes knowledge of causal mechanisms as that which is explanatory in scientific research. This method Salmon describes as the ontic conception of scientific explanation: Scientific explanation, according to the ontic conception, consists in exhibiting the phenomena-to-be-explained as occupying their places in the patterns and regularities which structure the world. Causal relations lie at the foundations of these patterns and regularizes; consequently, the ontic concezption has been elaborated as a causal conception of scientific explanation.4 Under Salmon’s description, explanation must be more than fitting observations under laws or regular patterns. It also requires knowledge of the causal mechanism that is involved. These explanations focus on describing the causal factors involved in natural events; by laying out these causal connections, we generate explanations. Salmon uses the example of knowledge of tides. In pre-Newtonian physics, there was an understanding of the regularity of tide movement, and that it had a relationship to the position and phases of the moon. Yet such accounts lacked explanatory value; only when " Salmon, Scientific Explanation and the Causal Structure of the World, 240. ‘2 Salmon, Scientific Explanation and the Causal Structure of the World, 239. 32 the at dema be no, he a to Salt explan still dis problem Shadow problem thereofi not Vice dissolvin H explanalit .iDOlIOm- L appropria- the account makes reference to gravitational pull do we move from ‘mere regularity’ to an ‘explanation’.43 One could ask, does the ontic conception require that a causal history be demanded as part of all scientific explanations? The answer, according to Salmon, would be no, although it is the primary strategy for scientific explanations. One alternative may be a “structural” explanation, which need not be causal.44 Such exceptions are suggestive to Salmon from the work of Peter Railton, yet Salmon maintains that many such explanations may still have underlying causal mechanisms that are as of yet unknown but still discoverable. A strength of Salmon’s mechanistic strategy is that it can handle some of the problems that arose for Hempel’s DN model, e. g., Jones’ birth control, the flagpole’s shadow, and the falling barometer. Salmon’s ontic explanation is able to address the problem of explanatory asymmetry by making reference to causal mechanism (or the lack thereof), thus showing that the flagpole length determines the length of the shadow amt not vice versa), and that the “blessing” of the salt had no relevant connection to its dissolving. How might Salmon’s mechanistic explanation and Kitcher’s unification explanation be related? Mechanistic models of explanation like Salmon’s are often called “bottom-up” theories: such explanation models begin with our causal understanding of basic, everyday phenomena. This understanding, when grouped and connected appropriately, provides us understanding of higher-level (or more abstract) phenomena. Salmon admits that his causal/mechanical and Kitcher’s unificationist explanations may be, in some ways, compatible. Calling his approach explanation] and a unification approach explanationz, Salmon argues the former may be at the heart of explanatory ‘3 Salmon, Scientific Explanation and the Causal Structure 0fthe World, 121. He also notes here that the ideal gas law is indeed law-like (that is, highly regular), but has very little explanatory value. 4‘ Salmon, Four Decades of Scientific Explanation, 165. 33 pro jet \Kidi unere expiar strateg explan. underst projects about local causation. From here, we develop generalizations about the world. With explanationz, the concern is more global, or ‘top—down.’45 This admission is interesting to note not because it strongly discredits either theory, but shows that explanatory projects may have multiple viable routes, rather than a single set explanatory strategy. Salmon’s focus on mechanism also reflects much of what occurs in biological explanations. For instance, many examples of infection and disease rely upon our understanding of how infectious agents act in the body. As Salmon writes: [T]o understand AIDS, we must deal with viruses and cells. To understand the transmission of traits from parents to offspring, we become involved with the structure of the DNA molecule. . .. When we try to construct causal explanations we are attempting to discover the mechanisms—often hidden mechanisms—that bring about the facts we seek to understand.46 Thus, the discovery of underlying causal mechanisms, for example, the discovery of DNA, allowed for unifying the theories of biology and chemistry. A significant risk remains in Salmon’s ontic model. Salmon may be correct in that most scientific projects seek to develop ontic explanations. So, what counts as a ‘good’ medical explanation will be determined in large part by the underlying structure of the world. Yet this move presupposes that the mechanisms involved are ultimately discoverable or identifiable. Without a causal mechanism identified, an explanation has not been generated. This leaves the burden on Salmon to show that all, or at least most, explanations of interest in science will have discoverable causal mechanisms. What does this hold for explanatory possibilities involving many medical conditions? Consider the case of J.S. and “unexplained” medical disorders. In cases like these where there is significant debate or disagreement about the underlying causal ‘5 Schaffner, Discovery and Explanation in Biology and Medicine, 295. ‘6 Salmon as quoted in Schaffner, Discovery and Explanation in Biology and Medicine, 292. 34 mecha debate. the hen lack of achieve be disCt mechan pOSsihle gather d explanat half till) \ exPlanet: factor of “'hl'flue question: I explanati bl‘is for “5063mm k“O‘i'ledt ”‘0” of El ciplanau x Basc , mechanism, a possible solution may involve hope that further research may resolve the debate. Yet that is to lay explanatory possibilities on future conditions. J.S. is a patient in the here-and-now, and her doctor is being asked to explain her illness. In such cases, the lack of understanding about the causal mechanism means that explanation has not been achieved. What about situations where there is little to no hope that any mechanism will be discovered, and very likely not in a timely way to help J .S. If the underlying mechanism ultimately is not discoverable, then explanations of that event would not be possible. This tension regarding time and explanations, that is, how long can we wait to gather data, is an important fact about types of medical explanations. Research explanations can afford the time to gather better data, while clinical explanations do not have this luxury. Van F raassen ’s Pragmatic Explanations Van Fraassen has provided significant work on the pragmatic features of explanations; his work promotes a fully pragmatic theory of explanation.47 An important factor of this explanatory model is the concept of ‘relevance’. Explanations, as answers to why-questions, must be relevant; relevance is determined by the context from which the questioner asks the why-question. This is in sharp contrast to, say, Salmon’s mechanistic theory of scientific explanation, where understanding the causal mechanism is the universal and acontextual basis for scientific explanations. Instead, van Fraassen argues that explanations are necessarily contextual. We ask why-questions from a certain context of background knowledge and assumptions. As van Fraassen describes, we use a “relevance relation”, a sort of filtering mechanism that determines which causal factors we utilize as explanatorily salient, and which causal factors we set aside. A good explanation will answer the why-question asked from this context, and the context may vary between ‘7 Bas C. van Fraassen, The Scientific Image (New York: Oxford University Press, 1980). 35 lndisi be a g relatio indepe explan relevar require Without inform: a theor} and app individuals. A good explanatory response developed with respect to one context may not be a good explanation with respect to another context. Because of the essential relationality of pragmatic explanations, van Fraassen rejects the possibility of an independent perspective from which to develop an explanation or from which an explanation may be judged explanatorily successful. Van Fraassen’s pragmatic explanation has been criticized strongly. Because relevance is determined by context, van Fraassen has argued against additional requirements. Salmon and Kitcher have argued that this leaves such a model vulnerable: without some defined limits to scientific relevance, it is possible one can find any and all information relevant, and thus explanatory. Still, I believe these problems are not fatal to a theory of pragmatic explanations. In Chapter 5, I return to van Fraassen’s explanations, and apply it to medical explanations. Schaffner’s Middle Range Explanations Kenneth Schaffner has argued for a different explanatory strategy in his detailed study of biological and medical explanations. This account of explanation in biology and medicine utilizes six explanatory components. These six components are: (E1) the Semantic (or biomedical system) Component; (E2) the Causation Component; (E3) the Unificatory Component; (E4) the Logical Component; (E5) the Comparative Evaluation Inductive Component; and (E6) the Ideal Explanatory Text Background Component.48 “The recurrent theme regarding explanation in this chapter has been the appeal to (possibly probabilistic) causal model systems which instantiate generalizations of both broad and narrow scope of application.”49 Here, Schaffner argues for the possibility of deterministic or probabilistic causation as explanatory in medicine (as opposed to teleological or functional explanations). 48 Schaffner, Discovery and Explanation in Biology and Medicine, 322. ‘9 Schaffner, Discovery and Explanation in Biology and Medicine, 322. 36 .-__4A-—« is arr as at biolo infon biolo; range and th Schaff Parlicu allow t often a] POPUIat Closer l( “Planut approac} medica} ' Ir. mOdel. \ which of guidance ”Wag: Want. Note that Schaffner’s approach is not a specific model, and as such its usefulness is ambivalent. One reading may be that this approach holds open 6 types of questions that as a whole are explanatory. Or, a more modest (but more practical) reading may be that biological and medical explanations often ask about (roughly) six various types of information. As such, these are different effective strategies that are often called upon in biological and medical explanations. Schaffner refers to biological and medical explanations as “theories of the middle range” since they fall midway between their use of ‘universal’ theories of biochemistry and the theories of evolution or of population genetics.50 Such middle-range theories, Schaffner argues, are applicable to explanations involving universal theories and unique particular findings. Middle range theories are useful in biology and medicine since they allow us to reason by analogy. By having a middle range of explanatory theories, we are often able to move broader (via emphasis on biochemistry) or more narrow (via use of population genetics) to explain either groups or individuals. Schaffner believes this is closer to biomedical explanatory practice, practice that does not reflect, say, deriving the explanations from covering laws in a Hempelian strategy. In this way, Schaffner’s approach may come closer to how doctors (and not just medical scientists) develop medical explanations. , In this way, Schaffner is right to say that it is too simplistic to give any one single model. What remains unclear, though, is how doctors determine (or should determine) which of these six explanatory strategies to employ at any given time. Also, there is no guidance on what to do when faced with possible conflicts: e. g., if there is an emphasis on pragmatic considerations (E6), must the medical explanation always involve a mechanism (E2)? 5° Schaffner, Discovery and Explanation in Biology and Medicine, 65. I again take up Schaffner’s concept of middle-range theories in Chapter 3 in discussing their relations to Paul Thagard’s work on explanations as Causal Network Instantiations. 37 Railton ’s Ideal and Non-ideal Explanations Peter Railton distinguishes two strategies of explanation theory.51 One begins with the logical structure of an explanation, and our strategies to meet such criteria. The other begins instead with our understanding of the language of explanations (philosophy of language), and what is meant or implied by explanatory requests. As Wesley Salmon describes Railton’s work, the “distinction between the ideal explanatory text and explanatory information can go a long way. . .in reconciling the views of the pragmatists and the realists.”52 This is important since ‘explanation’ can have different meanings and nuances. For example, ‘explanation’ can mean a portion of the explanation, as well as an ideal one. The distinction Railton makes is between ‘ideal explanatory texts’ and ‘explanatory inforrnation’. Railton believes that our explanatory requests often confuse these two types of information, but that they can and should be separated. Salmon, on the muddling of these two types of explanations, says, “One useful way to think about this conflict, I believe, is to regard the objectivists—the advocates of the ontic conception—as focusing on the ideal explanatory text.”53 The ideal explanatory text will be a complete description of all the relevant causal factors involved in explaining a given event. Ideal explanatory texts can be “expected—in most, if not all, cases—to be brutally large and complicated?“ While this ideal approach may have some advantages, it is important to ask if an explanation can be so ‘brutally large and complicated’ that no human can understand it, then has explanation really been achieved? If ideal explanatory texts were 5' Peter Railton began this work in his dissertation: Explaining Explanation. PhD. diss., Princeton University (1980). But see also his “Probability, Explanation, and Information” Synthese 48 (1981): 233- 356, and Railton’s “A Deductive-Nomological Model of Probabilistic Explanation” Philosophy of Science 45 (1978): 206-226. ’2 Salmon, Four Decades of Scientific Explanation, I61. ’3 Salmon, Four Decades of Scientific Explanation, 161. In this way, Hempel may also be an objectivist, although Salmon does not discuss him directly here. 5‘ Salmon, Four Decades of Scientific Explanation, 159. 38 our 0111) nor beer Explana an ideal explanat explanal I raised in (sometin [hey Cam achieve 1 Patient Cl read this consent. I 3 medic; Y] the Only t requeSl {C our only route to explanations, then I would say that in many situations explanation has not been achieved.55 Yet Railton allows for another explanatory option: explanatory information. Explanatory information could be the stuff that scientists gather together to get as close to an ideal explanatory text as is reasonably possible. Most actual requests for scientific explanation—most explanation-seeking why-questions—are requests, not for ideal explanatory texts, but, rather, for explanatory information. Let us note a relevant worry about clinical disease explanations, one commonly raised in bioethics discussions on informed consent. In brief, doctors often note (sometimes sarcastically, sometimes reflecting the pressure of the moment) that they feel they cannot teach the patient everything they need to know about their disease in order to achieve truly informed consent. The quip is often something like, “I can’t teach the patient everything. . .there’s a reason I spent four years in medical school.” One way to read this frustration is that the doctor feels that to live up to the rigors of truly informed consent, the patient would need access to an ideal explanation (or at least as close to one as medical science allows), and the patient would need to understand the explanation. Yet this move is problematic. First, it wrongly assumes that ideal explanations are the only beneficial route available. As Railton notes, not all requests for explanations are request for ideal explanatory texts: . . .even if we did possess the ability to fill out arbitrarily extensive bits of ideal explanatory texts, and in this sense thoroughly understood the phenomena in question, we would not always find it appropriate to provide even a moderate portion of the relevant ideal texts in response to particular why-questions. On the contrary, we would tailor the explanatory information provided in a given context to the needs of that context; if we had the capacity to supply arbitrarily large amounts of explanatory information, there would be no need to flaunt it.“ ’5 For my argument on this, see my discussion of Sandra Harding and the theme of complexity in Chapter 1. Although I will not argue for this in detail here, in brief I believe that if explanations are only possible in structures that human beings cannot grasp, then a key goal of explanation has failed. 5° Railton, “Probability, Explanation, and Information”, 244. 39 So ext neces: many often 5 may bl contex contex to med is an at situatic CXplan; llnfortu mfmma the Situ; iipeq o r 918 ‘i‘ant generate generate. like? u glVeS 1”, remains So even if the doctor could provide an ideal explanation to the patient, this is not necessarily what is called for in this situation. In this sense, van Fraassen is correct. In many cases, patient may only need a specific portion of the ideal text—that is, patients often seek explanatory information, rather than the entire ideal text. Clinicians, though, may be interested in different explanatory information. Thus, the ability to understand the context of the situation where the explanatory request is made, and how to respond to this context by crafting the best non-ideal explanation, is a skill that has practical importance to medical clinicians. Railton acknowledges this. What is important but missing, though, is an account for determining which explanatory information is appropriate in a given situation. Yet Railton does not give guidance for navigating the process of developing explanatory information for different audiences. In this way, there has been an unfortunate lack of attention to understanding the project of developing explanatory information that meets the needs of inquisitors, say, patients or physicians. As Salmon points out, “However, if it is only the ideal text that is so forbidding, the situation is not so hopeless. The ideal explanatory text contains all of the objective aspect of the explanation; it is not affected by pragmatic considerations. It contains all relevant considerations?” It may be the case that while ideal explanatory texts cannot be generated in a useful way to describe patients’ health, explanatory information can still be generated. But what would this explanatory information (or non-ideal explanation) look like? While Railton acknowledges that often non-ideal explanations are called for, he gives little to no guidance on how to construct these non-ideal explanations. Thus, it remains unclear how the pragmatic features will limit which features of an ideal explanatory text we use in providing non-ideal explanations, e.g., between doctors and patients. It is here that van Fraassen gives us more guidance. ’7 Salmon, Four Decades of Scientific Explanation, 161. 40 There is another concern raised by Railton’s ideal explanations. I worry here that Railton’s distinction places a greater importance upon “ideal explanatory texts” as the real or actual (and therefore meaningful) explanatory project.58 My concern here is whether clinicians, in explaining disease to patients, are actually picking out portions of an ideal text to provide to patients. Here, I agree with Salmon’s discussion about the contextuality of most explanations, and the relationship to ideal explanations: We must admit, in agreement with van Fraassen, that pragmatic or contextual factors play a large role in determining what sort of explanatory information is pertinent in any particular situation. The term ‘scientific explanation’ itself is ambiguous; sometimes it refers to some appropriate explanatory information supplied in a given context, but sometimes it refers to an ideal explanatory text. Under the first construction, there are such things as genuine scientific explanations; under the second, it may be that genuine scientific explanations represent ideals that may sometimes be approached but are (almost) never fully realized. Railton’s discussion contributes importantly to the clarification of the relationship among these concepts.59 As such, much of the work in explanation theory has focused on what would fulfill an ideal explanatory text, rather than the more practical (and more often more possible) project of investigating the nature of explanatory information. If medical explanations can be conceived of as either ‘ideals’ or as ‘explanatory information’, how would these be different? I agree with Salmon that most requests for both scientific and medical explanations are not requests for ideal explanatory texts. Instead, many are requests for explanatory information. If we understand ideal explanatory texts to hold all relevant information, what means do we have of whittling down these cumbersome creations in order to create something with meaning to patients? The answer to this, as Salmon and Railton acknowledge, will have much to do with 58 For example, I think Paul Thagard’s explanations as a detailed causal web—what he calls Causal Network Instantiation (CNI)—is an example of this prioritizing of ideal explanations, although he does not use Railton’s language. I introduce and critique Thagard’s CNI theory of medical explanation in detail in Chapter 3. ’9 Salmon, Scientific Explanation and the Causal Structure of the World, 264. 41 ex ex; atte imp SUpe understanding the context of patients’ situation and other pragmatic factors, but they provide little guidance on how to generate responses to why-questions seeking explanatory information. One way of thinking about the relationship between ideal and non-ideal explanations is suggested by Salmon’s and Railton’s language. It may be that non-ideal explanations are portions (or subsets) of ideal explanations that we choose to call attention to. This implies that certain laws, facts, and mechanisms may be of greater importance for a given context. The rest of the ideal explanation, although accurate, is superfluous in this context. While this is one possible relationship between ideal and non-ideal explanations, it has a number of limitations. First, it implies that pragmatic considerations are involved only in determining which subset of the ideal is explanatory in a given context. Second, it implies that the ideal explanatory texts may stand alone, yet explanatory information is always dependant upon its relationship to the larger ideal text. The concept of ideal is meant to be in some sense objective: ideal explanations contain all information that is explanatorily relevant. But Salmon and Railton do not criticize this conception of ‘ideal’. For instance, they do not ask the question of “Ideal for whom?” They may mean to equate ‘ideal’ with ‘complete’. But even if this is the case, complete information may vary between different scientists, not to mention between a patient and a clinician. Third, the non-ideal explanation is structurally dependent upon the ideal explanation. Just as Hempel noted that we often provide ‘explanatory sketches’f’0 that could be expanded (should we choose to invest the time and energy) into proper DN explanations, so too non-ideal explanations are imperfect, serving as substitutions for ideal explanations. 6° For instance, Hempel notes in one example that we can say that the butter melts because the pan is hot, but not yet have provided any of the laws of thermodynamics involved. 42 depent ideal e occurs believe cousin complc‘ magma factors AS S‘dlr lane ()1 lJLSIEad. nOil-ide Consider i PTOCegS‘ comm. may int. explanar which is though Heparin Ciplana Fourth, the misunderstanding of non-ideal explanations to be structurally dependent upon ideal explanations has led to other problems. This view has allowed non- ideal explanations to be seen as “dumbed down” versions of ideal explanations. This occurs, for example, when clinicians provide only parts of the explanation that they believe their patients will be able to understand. Instead, non-ideal explanations can be creations that are independent of their ideal cousins. The pragrnatics of generating non-ideal explanations may be more interestingly complex than is implied by Salmon and Railton. It may be that in some circumstances, pragmatic considerations guide our attention to explaining to a patient the environmental factors involved in her developing cancer (rather than, say, her genetic predisposition). As Salmon and Railton, imply this is to discuss only certain threads of the ideal, and to leave other threads unmentioned or unconsidered.61 Yet not all non-ideal explanations are such limited or partitioned creations. Instead, some causal threads are regrouped, redescribed, or summarized as part of the non-ideal explanation. Pragmatics may force us to refer to other factors that had not been considered as part of the original ideal text. So, the generation of non-ideal explanations can sometimes be a subtractive process, beginning with the ideal and taking away features that are unnecessary for this context. Yet pragrnatics may also be involved in the addition of other information. This may involve reference to information that is not immediately relevant, but is still explanatory to the patient. Consider that Frank has been diagnosed with Hepatitis A, which is typically spread through contaminated food or water. It may be explanatory, though, for him to know that his condition is different from his friend who contracted Hepatitis B, which is typically caused by exposure to blood or by sexual contact. The explanation of Hepatitis B is not part of the ideal explanation of Hepatitis A. Yet to 6‘ Note here I am switching metaphors from ‘explanatory texts’ to ‘causal webs’. 43 Frank of the Con prelim explan chapter CKplan. general mechat Show, t In Char Frank, it may be important to provide this additional explanation. (I will take up this issue of the relationship between ideal and non-ideal explanations again in the next chapter.) Conclusion Thus ends a brief overview of proposed theories of scientific explanations. This preliminary discussion is sufficient at this point to begin a conversation about medical explanations and the benefit of explanation theories previously proposed. In the next chapter, I explicate Paul Thagard’s theory of medical explanation. Successful medical explanations for Thagard rely upon detailed causal knowledge of how diseases occur in general in human beings. In this way, a detailed knowledge of disease causation and mechanism is utilized much like the explanatory theories of Salmon and Railton. I will show, though, that such explanatory approaches are problematic for a number of reasons. In Chapter 5, I utilize van Fraassen’s pragmatic explanations to rectify these problems. CHAPTER 3: A CRITIQUE OF THAGARD’S THEORY OF MEDICAL EXPLANATION S AS CAUSAL NETWORK INSTANT IATION Introduction Given Peter Railton’s distinction between ideal and non-ideal explanations that was described in the previous chapter, I continue to question the proper goals and structure of medical explanations.62 I begin by examining the explanatory schema articulated by Paul Thagard in How Scientists Explain Disease: explanations as causal network instantiations (CNI). Thagard intends his work to illustrate how explanations of disease are generated and evaluated?3 Such explanations are utilized by both biomedical researchers and individual clinicians. After articulating Thagard’s CNI, I conclude that such an account is problematic for a number of reasons, and thus it is inadequate as a basis for medical explanations. Thagard’s explanatory schema places significant emphasis on showing how phenomena fit into a causal nexus. In this way, Thagard conceives of medical explanations—both clinical and research explanations—as what Salmon calls ontic approaches of explanation. As Salmon writes, to give an ontic explanation is “to show how events. . .fit into the causal structure of the world.”64 Although promising and beneficial within limited contexts, Thagard’s explanation as CNI cannot adequately serve as a basis for medical explanations generated as part of clinician-patient discussions. Rather than advocate an ontic conception of medical explanation, I argue that medical explanations are best conceived as erotetic approaches. That is, in brief, medical ‘2 Railton, Explaining Explanation, and “Probability, Explanation, and Information”. 63 Thagard has a secondary goal in the overall book, that of understanding how advances in medical science are possible. Although interesting, I do not take up this line of discussion in my work here. This does suggest, though, something about how Thagard’s and my perspectives differ on approaching an analysis of medical explanations. This may also go towards explaining why he has either overlooked or chosen not to take up some of the problems in which I am most interested. 6" Wesley Salmon, Scientific Explanation and the Causal Structure of the World, 19. 45 explanat Salmon knowled intellect relativiz explanal cxplana: l knowlet disease “clinica tXplana “Plans address bener e Pallemh of 3&3 , baSlS f0 explanations ought to answer patients’ why-questions about their health. According to Salmon’s description of erotetic explanations, such explanations “fill a gap in someone’s knowledge; the adequacy of an explanation is judged in terms of the manner in which the intellectual lacuna is filled.”65 One implication of this is that “explanations must be relativized to knowledge situations.” 66 I draw upon the conception of erotetic explanations to argue that medical explanations should be crafted for the needs of explanatory inquisitors. While explanations generated as part of biomedical research may be relativized to knowledge-situations, the focus of most of my discussion will be on explanations of disease generated in doctor-patient conversations. Such explanations, which I call “clinical explanations”, should be crafted to meet the needs of the patient(s) for which the explanations were constructed. In this chapter, I begin by clarifying the features of explanation that are important to medical explanations, but that Thagard’s CNI cannot address. In the following chapters, I will advance a model for medical explanations that is better equipped theoretically to emphasize the importance of these factors in addressing patients’ why-questions. Specifically, in Chapter 5 I will utilize the explanatory structure of Bas van Fraassen—who has provided a detailed, erotetic explanatory model—as the basis for medical explanations that can meet patients’ explanatory needs. Before providing an alternative model, though, I will first attempt to provide a characterization of such an ontic approach to medical explanations (that of Paul Thagard), and then discuss it critically to show why it remains an insufficient schema as the basis for medical explanations. In this chapter, I argue Thagard’s CNI falls pray to the myth of the ‘ideal explanation’ (a la Railton and Salmon): medical explanations are assumed to begin as an ideal and static causal web of all possible factors that are relevant. Such an approach ‘5 Salmon, Scientific Explanation and the Causal Structure of the World, 101. 6” Salmon, Scientific Explanation and the Causal Structure of the World, 101. 46 prohibi hand.\ approp: meho merek those tl exrdan fails to genera explan femox' llnrd, Episte C(mch them-1 llec file a; mem- PTact prohibits viewing the explanation as relativized at the outset to the patient or situation at hand. Yet Thagard’s strategy for generating non-ideal explanations that would be appropriate for patients’ why-questions is inadequate. First, Thagard mistakenly limits the broad range of work done by medical explanations, and he improperly characterizes the relationship between medical explanations that are part of biomedical research and those that are part of doctor-patient discussions. Thus, he wrongly sees medical explanations generated for patients as a part of ideal explanations. Second, this strategy fails to take seriously the difficulties of the pragmatics of explanation involved in generating and evaluating these clinical explanations”, and assumes that generating explanations for patients is only a matter of whittling down ideal explanations, that is, removing information from that ideal explanation that is irrelevant to the case at hand. Third, Thagard’s explanatory strategy wrongly ignores patients’ status as authoritative epistemic agents who participate in the generation of their medical explanations. I will conclude by arguing further that these problems should be remedied by improving a theory of medical explanation to better respond to patients’ why-questions. Medical Explanations as Clinical and Research Explanations In his book How Scientists Explain Disease, Paul Thagard details how diseases are explained in medical research and practice. His goal is to develop an explanatory theory of medicine that reflects the actual practices of biomedical research and of clinical practice. Early in his book, Thagard divides medical explanations roughly into two categories: Two kinds of explanation are important in medicine. When a patient goes to a physician with a set of complaints and symptoms, the physician’s first task is to make a diagnosis of a disease that explains the symptoms. For example, if the ‘7 As I noted in Chapter 1, I am not dealing directly with standard discussions on the discovery and justification distinction. Thagard, though, is more interested in some of these themes, which for the sake of my project I do not take up in a strong sense. See his “Chapter 3: Ulcers and Bacteria: Discovery” and “Chapter 8: Discovering Causes: Scurvy, Mad Cow Disease, AIDS, and Chronic Fatigue Syndrome” in How Scientists Explain Disease. 47 luil “clin this c thisc Noe Clinic resent CXplct “plat CXplaj patient has a fever, muscle aches, and a runny nose, the physician may explain these symptoms by saying that the patient has influenza. The second kind of explanation, which belongs to medical research rather than clinical practice, requires an answer to the question of why the patient became sick with influenza, which we now know is caused by a virus.‘58 I will call these types of medical explanations (using my own terms, not Thagard’s) “clinical explanations” and “research explanations” respectively. In the first portion of this chapter I develop this distinction. I then return to these same concepts to show that this distinction is not as clear as it initially appears, and the relationship between these two explanatory projects is not as straightforward as Thagard remarks. I then argue that clinical explanations do a wider range of work than is implied initially by Thagard. Research explanations are exemplars developed by clinicians and biomedical researchers to explain how a disease typically works in the human body. Research explanations come closest to biological explanations (as opposed to scientific explanations generated in other domains). Indeed, research explanations are biological explanations in the specific context of human health and disease. For instance, a research explanation can be generated to answer, “Why does the influenza virus cause sickness in the human body?” or “How does HIV respond in the body such that it results in AIDS?” The focus of these explanations is rather wide and general; these do not consider the specifics of any one individual, but instead try to explain how disease occurs generally within humans. They are abstractions away from a particular case. In some cases, they are explanations about why members of a certain group are more likely to become ill.69 Research explanations can also seek to clarify more distant causal factors of disease. If a virus is determined to be the cause of an illness, researchers may want to research how the infection became possible. Are there contributing factors that make 68 Paul Thagard, How Scientists Explain Disease, 20. ‘9 Thagard discusses why people in underdeveloped countries are more likely to have gastritis than are North Americans, and why nuns are more likely to get breast cancer than are other women (Thagard, How Scientists Explain Disease, p 116). I raise the discussion of explanation of disease for groups later in this chapter, and again consider Thagard’s examples. 48 infection more (or less) likely? In this way, research explanations investigate more distal causal factors than may be considered as part of clinical explanations.7o We can look also at research explanations from a historical point of view, and understand them as trends or research patterns that medical researchers have followed. As such, these trends have changed over time, and have simultaneously directed the questions raised by medical research. This analysis is one that is particularly of interest to Thagard, and it is directly responsible for his consideration of the explanation of the gastric ulcers as a theme threaded throughout his book. This disease, Thagard notes, is particularly interesting because medicine has had numerous explanations for the cause of ulcer formation over time. While such historical considerations of research explanations are interesting, these are, for the most part, outside what I am considering. The one exception may be that historical changes in medical explanations do tend to linger, especially in the lay public’s understanding of a disease. Thus, some people, rather than seeking a doctor’s treatment, may attempt to cure their own ulcers by drinking a bland diet (e.g., lots of milk, little spicy or fatty food). While such remedies may have been prescribed by clinicians in the past, this dietary solution has been found ineffective. Clinical explanations are narrower in scope; often by applying portions of research explanations, clinical explanations are developed to explain the sickness of an individual. Thagard suggests that a primary function of physicians is to provide diagnoses, that is, to explain why a patient exhibits this set of symptoms. For instance, if Jane displays a fever, congestion, and general body aches for a number of days, her physician may diagnose her with the flu. Thus, the task of providing diagnoses is directly related to clinicians’ ability to craft clinical explanations. 7° I will return to this point, arguing that clinical explanations too often look at proximal, rather than distal, causes of disease as explanatory. Contextual and pragmatic features, I will argue, determine the proper “range” of causal factors that should be considered as explanatory. 49 One problem with this view, however, is that Thagard asserts without arguing for his assertion that clinical explanations are always portions of research explanations, rather than things generated independently from research explanations. I suggest that the relationship between research explanations and clinical explanations is more complex than Thagard’s reductive scenario can allow. While clinical explanations may draw upon research explanations for information, clinical explanations are structured differently (because they are based on different why-questions). I return to this problem later in this chapter. Reevaluating the Clinical/Research Distinction Thagard’s original binary of medical explanations—clinical and research—is too simplistic. Although it informs the explanatory schema Thagard ultimately argues for, which will be discussed in the following sections, I want to point out some of the complexity around medical explanations that I think Thagard has overlooked. Once uncovered, understanding this complexity of the work done by medical explanations will inform my later discussion and criticisms of Thagard’s explanatory schema. There can be a wide range of goals in medical explanations. Below I describe some rough types of explanations that can be requested. This is not a complete list, nor are these categories cleanly separable from each other. Nor are any these explanatory requests unique to either clinical nor research explanations (although some may be more common in some contexts than others). What all medical explanation share is a sense that understanding ought to be improved. What varies between categories are the topics involved, the agents requesting and receiving the explanation, and the context in which the explanation is generated. Sometimes this work is meant to benefit the clinician’s work; other explanations are generated to benefit patients’ understanding. Again, there is rarely a neat division available here. It may be that ‘clinical explanations’ and ‘research explanations’ are closer to family resemblance concepts, since these are not easily 50 det has: IDEC rese care disc: to) b resea these resea- Panic Ciplm Hemp Hemp ahead; defined nor are necessary and sufficient conditions available. These are not categories based upon the structure of the explanation, but instead upon what the inquisitor wants to receive from her explanatory request. Having laid out here different types of work that medical explanations may do, I will distinguish in greater detail between clinical and research explanations. Again, this is not a firm distinction, but a generalization about both categories. Research Explanations Research explanations are epistemic projects that try to make clear how a given disease works in the human body. In many ways, these can overlap with (or be identical to) biological explanations that focus upon human health and disease. Biomedical researchers generate research explanations and refine them over periods of time. Because these are explanations of general categories of disease or about a specific disease, research explanations do not focus upon the particulars of how a disease works in a particular person. Research explanations are generated in more abstract terms. Research explanations are typically highly detailed in scientific information, or at least like Hempel’s explanatory sketches, they could be made clearer by filling in the details. As Hempel noted, such details are often not explicitly stated." We believe the audience already knows these facts (or enough of these facts) such that they can fill in the proper information. If they do not know the facts already, they are somehow basic enough so as to be easily learnable. Such ‘gappy’ explanations are often provided as part of medical explanations: a scientist or the clinician, as explainer, may generally assume listeners have enough of a scientific background to fill in the gaps properly (with their background information or a bit of reading up on the literature).72 As such, research explanations are what I call peer-peer explanations: they are generated by clinicians or researchers for use 7‘ Carl Hempel, “Aspects of Scientific Explanation” in Carl Hempel, Aspects of Scientific Explanation and Other Essays in the Philosophy of Science (New York: Free Press, 1965) 331-496. 72 “Gappy” is a term borrowed here from Mackie to discuss his INUS conditions. See Schaffner, Discovery and Explanation in Biology and Medicine, 301-302. 51 by other clinicians and researchers. Explanations derived directly from biomedical research will research explanations. Similarly, even explanations generated by two clinicians about the care of a patient—that is, generated by peers—will be a research explanation. Clinical Explanations Clinical explanations, though, are usually non-peer explanations: clinicians generate these explanations for patients’ use, typically as part of an informed decision- making process about patients’ health. Clinical explanations are interesting here due to their importance as both epistemic projects and moral/political projects. Rather than generated as part of a lab or research setting, clinical explanations are most frequently generated as part of clinical encounters, that is, during the face-to-face communication between doctors and patients. For instance, diagnosis is often a clinical explanation developed on the spot during a clinical visit. Again, explanations are tools that are constructed for the job at hand. In this way, the moral/political importance of clinical explanations is closer to the surface.73 In many ways, I do not think that a highly (scientifically) detailed explanation is called for as part of many clinical explanations. Biological explanations and research explanations, which involve the human body and human diseases, are about increasing our understanding of natural phenomena. Clinical explanations instead have the goal of increasing our manipulative and predictive abilities.74 For instance, it is often 73 This is not to say that research explanations are apolitical or non-political. In Chapter 6, I will explicitly take up the question of how explanation theory in medicine, involving both clinical and research topics, has political interests shaping explanatory responses to why-questions. To do this, I will look closely at the works of feminist philosophers of science, including Sandra Harding and Helen Longino. 7" Salmon makes roughly this point in his discussion of two divergent explications of a single event (a plane crash due to ice on the wings). FAA regulators may want to prevent future crashes; a scientist may want to gain a deeper scientific understanding of the air flow and drag. Salmon writes, “Perhaps, then, there are two different sorts of explanation: explanation that increases our manipulative and predictive abilities, and explanation that increases our scientific understanding of natural phenomena.” Salmon continues, saying, “I suspect that van Fraassen has succeeded admirably in capturing the first of these,” that is, explanation as improving manipulative and predictive abilities (Scientific Explanation and the Causal Structure of the Universe, 134). Salmon worries, though, that van Fraassen’s explanatory account fails to account for a 52 unnecessary to have a complete causal explanation of how Jane contracted the flu in order to treat effectively her flu.75 Thus, Thagard’s CN I fails to reflect the actual explanatory practice structuring doctor-patient conversations of disease. The setting in which clinical explanations are generated determines the resources available. Often, research explanations will be utilized as background knowledge. There are practical limits on the amount of time that one can gather (possibly) explanatory information. This information may be ultimately unavailable, or the effort and resources that would be needed make gathering this data unlikely. Because these are generated as part of clinical encounters, there may often be a sense of urgency to shorten (rather than extend) the time taken to gather data and generate an explanation. The value of acting quickly to treat when, say, examining a patient in the ER, may outweigh the value of a slow, reflective process to generate a possibly better explanation over time. Clinical explanations, though, go beyond purely epistemic concerns; they are also moral—political projects. If clinical explanations are meant to improve patients’ understanding of disease, then if the explanation has been successful the patient should understand it. At the most basic level, all decisions to inform patients (or to withhold scientific understanding of natural phenomena, which Salmon argues his own explanatory theory can do. The point to be considered for this discussion, though, is that both research and clinical explanations improve manipulative and predictive abilities, but clinical explanations do this in a much more immediate fashion. Clinical explanations that do not improve patients’ ability to predict and manipulate their health conditions are failed explanatory attempts. 7’ A detailed causal explanation, of course, may have been influential in developing effective strategies, say, at the level of drug development. But many practices in medicine have become routine because they are effective, not because we have a deep causal understanding of why they are effective. For example, consider the history of puerperal fever, an epidemic hitting women after childbirth throughout the mid-nineteenth century. During this epidemic, which was before germ theory had been developed, the cause of the illness was unknown. In hindsight, we now understand puerperal fever as resulting from physicians—often coming from surgeries or autopsies to deliver babies—transmitting bacteria from their hands to women during delivery. Yet a well-developed “germ theory” was not necessary to explain that physicians rather than midwives (who did not engage in surgery) were spreading the fever by touch, or that antiseptic techniques could help curb spread of this epidemic (this was determined by scientific trials). In such examples, it may be possible to fill in the details of causation. But a level of explanation, understanding, and therapeutic response were possible without an understanding of deeper causal mechanisms. For a discussion of the history of puerperal fever, see Richard Wertz and Dorothy Wertz, Lying-In: A History of Childbirth in America (New Haven: Yale University Press, 1989) 1 12-128. 53 information, say, from a dying child) are moral decisions. Clinical explanations are crucial for generating informed, autonomous patient decision-making. If the explanation is presented in a format making it inaccessible to the patient, then the explanatory process has failed. Here, I am assuming the explanation is correct, and that another doctor could have understood the explanation. But in such a case, a research explanation, not a clinical explanation, was developed, and this may be inappropriate for the context of discussions with patients. It may also be the case that some research explanations are so complicated (or draw from different areas of science) that other clinicians fail to grasp them. I believe it is reasonable to assume in such cases that clinicians typically have means for further investigation to help them understand the explanation. This strategy is often unavailable to most patients as scientific laypersons. Clinical explanations are also more complicated in that they are shaped by additional pragmatic factors of the clinical encounter (factors that may or may not be epistemic considerations). Thagard’s Rudimentary Explanation Schema Having shown that medical explanations are called upon for a number of different types of work, we can now ask whether structural similarities exist in medical explanation responses. Thagard provides an initial attempt by noting that medical explanations are often grouped by disease types. Underlying these grouped disease explanations are similarities in the causal factors cited. Thagard argues first for what he calls a basic “disease explanation schema”: 54 Explanation target: Why does a patient have a disease with associated symptoms? Explanatory pattern: The patient is or has been subject to causal factors. The causal factors produce the disease and symptoms.76 Under this basic explanation schema, Thagard argues that the boldface words above can be replaced with more specific terms, those of individual patients, their environmental conditions, or specific diseases. Our explanatory interests shape which of these terms we expand upon and to what level of specificity. Explanation schemas consist of “an explanation target, which is a question to be answered, and an explanatory pattern, which provides a general way of answering the question.”77 These can be created at either the general or abstract level (i.e., research explanations), or with specific examples of a particular case (i.e., clinical explanations). Similar to other explanatory approaches, Thagard’s disease explanation schema begins with a why-question about the patient’s health. The explanation is the answer generated in response to this question. The explanation target is a why-question about the explanandum; the explanatory pattern is the group of sentences comprising the explanans. To see how this disease explanation schema works, we can consider a more detailed example, that of a multifactorial disease: cancer. While the basic schema pattern remains the same, we fill in the proper terms with greater details. Although this schema is written out as a textual explanation, it could also be mapped out as a causal diagram: Explanation target: Why does a patient get a cancer? Explanatory pattern: The patient has cells with active oncogenes resulting from a viral infection or a mutation of proto-oncogenes. These cells also contain mutated tumor suppressor genes. The tumor suppressor genes have failed to stop the stimulation of growth in the cells produced by the oncogenes, generating the patient’s cancer.78 7° Thagard, How Scientists Explain Disease, 20-21. 77 Thagard, How Scientists Explain Disease, 5. 78 Thagard, How Scientists Explain Disease, 32-34. 55 Her 10v exa this qne Cat in CC Cl (C7 Here, Thagard writes that our why-question (which is shaped by our interests) determines to what level of detail we fill in the bolded terms. (Unfortunately, though, Thagard provides no clear guidance on how to properly “fill in” the terms.) This example is presented at a level such that it can explain the development of cancers of various tissue types. It could be specified further to describe the development of cancer for a specific tissue type, or a cancer caused by a genetic mutation resulting from exposure to a specific toxin. This basic schema provides the structure for an explanatory response in this example to the why-question about cancer formation in an average, anonymous patient; this is an adequate response to this why-question. It would be possible to change the why- question to ask about the cause of a specific person’s cancer. In that case the appropriate causal details about a specific patient could be filled in (if such knowledge about this person is available). Note that this move from general underlying pattern of explaining disease in humans to explaining disease in a specific person makes it relevant for a clinical practice. Clinicians can begin by using the basic explanatory schema, and then add detail or specify a given concept, e.g., the number of genes at work, the specific tissue type, the type of virus. For Thagard, all that is required to explain the disease of a person is to fill in the blanks (or to replace the bolded terms) with the right amount of detail. This pattern comes back in his more advanced explanatory schemas, which I will discuss later in this chapter. For now it should be noted that Thagard implies a proto-pragmatic theory here, although Thagard does not identify it as such. Although his explanatory focus is on causal factors, he notes that the basic disease explanation schema requires us to fill in certain terms with the appropriate level of detail. What is missing, though, is an analysis of how a proper level of detail is determined. Thagard’s examples imply that researchers 56 can go back along causal pathways to varying degrees and consider a range of causal factors. He also implies that clinicians (with actual patients) will determine how far back along causal chains they ought to investigate. Clinicians also will investigate some types of causal features more closely, and ignore (or set aside) others. The relationship remains unarticulated between these contextual (or pragmatic) features of explanation and the ontic nature of Thagard’s CN I. By not addressing the former, he unfairly minimizes their importance.79 Hierarchical Organization of Disease Explanations Medicine utilizes a wide variety of standard explanation schemas, and as a group these allow for clinicians to explain a wide and diverse range of disease. When taken as a whole, Thagard argues this group of explanatory schemas is used as a categorization system for “disease” as a “hierarchical organization of disease explanations.”80 This system as a whole provides the closest thing medicine has to a unifying theory, equivalent to biology’s reliance upon evolutionary theory and molecular genetics.81 We can explain disease at different levels. We can explain Frank’s prostate cancer, we can explain how prostate cancer in general works, or we can explain how multifactorial diseases (of which cancer is one) work. We organize these middle-range explanations (the level of particular diseases) according to their causal mechanisms.82 Below is Thagard’s list of the hierarchical organization of disease explanations in medicine, focusing on this middle-range of explanations. The list is expanded somewhat 79 In Chapter 5 I will argue that pragmatic features of explanation are crucial to clinical explanations. 8° Thagard, How Scientists Explain Disease, 34-35. 8' Here, it is important to note that this hierarchy cannot be expressed in terms of an overarching law, the way biological theories can portray evolutionary theory and genetics. See Schaffner, Discovery and Explanation in Biology and Medicine, Chapter 6 “Explanation and Causation in Biology and Medicine: General Considerations”. 82 Although not specifically cited as such, Thagard’s use of the term “middle—range theories” seems to echo Schaffner’s work. In other places, though, Thagard explicitly cites Schaffner’s influence on his work. See Thagard, How Scientists Explain Disease, 34-35, and Schaffner, Discovery and Explanation in Biology and Medicine. p 65. 57 from the original figure, which Thagard notes is “partial”, using Thagard’s and my own examples.83 I Infectious 0 Bacterial, e.g., Lyme disease, tuberculosis, ulcers (from Helicobacter pylori infection) Viral, e.g., influenza, rabies Protozoa, e. g., cryptosporidiosis Fungal o Prion, e.g., Creutzfeldt-Jakob ' Nutritional, e.g., beriberi, scurvy, rickets I Molecular-genetics o Mendilian, e.g., cystic fibrosis 0 Multifactorial, e. g., lung cancer (emphasizing genetic and environmental interactions) - Autoimmune disease, e.g., lupus, rheumatoid arthritis 000 When taken together, infectious, nutritional, molecular- genetic, and autoimmune diseases comprise a significant portion of the disease topics covered by medical explanations.84 Thagard describes these explanatory schemas as “hierarchical” because of the different levels of groupings. General medical explanation of disease is comprised of a number of sub-categories: infectious, nutritional, molecular- genetic, and autoimmune diseases. Each of these categories, though, can be broken down further. For example, ‘infectious disease’ is comprised of various types of infectious agents, e. g., bacteria, viruses, or prions. Moving one level down this hierarchy, bacterial diseases can be broken down into types of bacterial infections, like Lyme disease, TB, Helicobacter pylori infection (which is the predominant cause of gastric ulcer formation). One advantage of this explanatory grouping is that it reflects basic intuitions about medical explanation. By grouping explanations according to similar causal mechanisms, we can roughly explain these diseases as working in similar ways. Before it was known that bacteria could survive the acidic environment of the stomach, a bacterial explanation of ulcers made little sense. We did know, though, that bacterial infections 83 Thagard, How Scientists Explain Disease, 35, Figure 2.9. 3‘ Of course, this list can be added to or edited as is warranted by ongoing biomedical discovery. 58 could can more to t explanatt knowledg (that is. r prototype they are r T. biomedic bl dil di: er could cause sores. Upon the discovery of H. pylori existing in the stomach, it was an easy move to explain ulcers as the result of bacterial infection. Here, Thagard’s basic explanatory schema works in similar ways to Schaffner’s “disease exemplars”, such that knowledge at a given level lets us analogize to similarities of mechanism for lower-level (that is, more specific) diseases of the same type. These exemplars are used as (interlevel) prototypes to organize information about other similar (overlapping) models to which they are related by analogical reasoning rather than deductive elaboration.85 Thagard relies upon explanatory unification, utilizing Schaffner’s ‘middle-range’ biomedical theories. Medicine generally explains disease: by fitting particular diseases into general concepts based on common mechanisms. Medical explanation is a matter of fit with causal schemas at different levels of generality, ranging from particular patients to particular diseases to levels of kinds of diseases. Unification in medicine is simultaneously explanatory and conceptual, because what ties explanations together is an organized system of disease concepts.86 At the highest of these levels, we have an explanation of how disease works. As Thagard continues, “In medicine. . .unified understanding does not come from the availability of a general overarching theory but from the availability of a system of explanation schemas?” This observation fits well with Kitcher’s unification theory of explanation in that explanations involve understanding at the level of patterns, rather than focusing on individual cases. Once we have these hierarchical organizations of disease explanations (a product of medical research), they help us to organize the causal relationships responsible for disease and symptoms within a given patient.88 This hierarchical organization of disease 85 Kenneth F. Shaffner, “Exemplar Reasoning about Biological Models and Diseases: A Relation between the Philosophy of Medicine and Philosophy of Science,” Journal of Medicine and Philosophy, 11 (1986): 63-80. 86 Thagard, How Scientists Explain Disease, 34-35. 87 Thagard, How Scientists Explain Disease, 34. 88 Thagard, How Scientists Explain Disease, 36. 59 ‘?m‘ A... 1 V erplanati grouping explanati in mm therefore clinical u the struct Medic B; additiona; causation dflinition interacting “cause“ is dhéase) ll explanation does much to organize groups of explanations that work similarly, but this grouping does not help us to better structure medical explanations. The basic disease explanation schema Thagard articulates, although helpful in a general way, is unhelpful in making fine distinctions within the explanation. I agree with Thagard here that it therefore has limited benefit as a basis for developing medical explanations, whether for clinical or research purposes. Because of these limitations, Thagard develops a theory of the structure for medical explanations: explanations as causal network instantiations. Medical Explanations as Causal Network Instantiations Before describing the details of Thagard’s preferred explanatory schema, two additional background points require clarification. The first involves a brief comment on causation and how I have been using the term. Generally, I do not employ a technical definition of “cause”, only a commonsensical consideration of contributing and interacting factors. For instance, I do not think a deeply philosophical conception of “cause” is implied by most clinicians in their discussions of etiology (the causes of disease), nor do the arguments I make about medical explanations in this paper require an in-depth philosophical analysis. Oftentimes medical causes seem to imply counterfactual conditions, such that if cause, had not been present, then disease, would not have occurred. Explanations involving causes often additionally require an overarching scientific law (law,) that connects with cause, to produce disease,, an explanation that cannot occur by citing the effect (disease,) alone. Again, looking at causation as involving universal laws raises a number of deep philosophical questions, but their resolution is not necessary here for my project. Thagard himself minimized his reliance upon a clear conception of causation for a successful explanatory project, noting that he has “not attempted to define cause in terms of explanations or explanations in terms of cause. Causes, mechanisms, explanations, and 60 explanat causes E di ferenl Under e well wit lnl’OlVC( laws ant the belie Without l 0i medic tiplititi eliminate ll‘hich ex lagme w accOunts boll] Clini E. develop,x many Pet do dwelt, explanatory coherence are intertwined notions.”89 For Thagard, when we say that C causes E, the causal mechanism is tied to “explanatory coherence”.90 Note that this is different from Salmon’s stricter reliance upon causal mechanism for explanatory success. Under explanatory coherence, the mechanism can be known or assumed, but must work well with other higher-level explanations. If we ask “Why E?”, the general mechanism involved in the explanation “C causes E” should make sense given other related causal laws and other accepted examples in medicine. What both Thagard and I share, though, is the belief that a clear and useful understanding of medical explanation can be developed without requiring we first resolve all philosophical debates about causation. The second background point involves locating Thagard in the larger discussion of medical explanations. Thagard develops his theory of medical explanation as an explicit alternative to other proposed explanatory theories. As Thagard writes, “we can eliminate a number of defective alternative accounts of explanation, including accounts in which explanation is essentially deductive, statistical, or involves single causes.”91 Here, I agree with Thagard’s main arguments, and I am sympathetic to his feeling that standard accounts of scientific explanation are inadequate as the basis for medical explanations—— both clinical and research explanations. Explanation in medicine is not deductive: although deductive explanations may be developed in other fields, they are rarely of the type sought in medicine. For example, many people who smoke do not develop lung cancer, and many people who do not smoke do develop lung cancer. While we know that smoking is causally connected to lung cancer, we cannot provide an overarching law about this connection. 89 Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms”, FN 11. This article later became Chapter 7 in How Scientists Explain Disease. 9° For a detailed account of Thagard’s understanding of causation—which for him is tied to explanatory coherence—see Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms”, FN 7, and Thagard’s How Scientists Explain Disease, 65 and 109-111. 9' Thagard, How Scientists Explain Disease, 113. 61 J utilized suificie too Sim Althout disease: bodily ( heritabl be mult. preferre netumk SilalEgv lll dlSCag dllCage f Explanation in medicine is not statistical: Although statistical information is often utilized as part of medical explanations, statistical information alone is not explanatorily sufficient for the needs of medical practice. Explanation in medicine is not in terms of single causes: As Thagard notes, it is too simplistic to describe the development of lung cancer as only the result of smoking. Although smoking may be a predominant cause, it is not the only factor. As in most diseases, there will be a wide range of environmental factors (e.g., exposure to toxins), bodily conditions (e. g., obesity or high blood pressure), and genetic factors (e.g., heritable traits or cellular mutations). In this way, explanations of disease will typically be multifactorial explanations.92 These points are background arguments to Thagard’s preferred view that medical explanations are causal network instantiations. Thagard argues that explanations of disease are best understood as an intricate network of causal factors, which he calls “causal network instantiations” (CNI). This strategy emphasizes the importance of detailed causal factors and mechanisms that result in disease.93 We gain our understanding of these causal factors involved in a given disease from biological research, epidemiological studies, and other sources.94 As we gather and refine our understanding of the causal factors involved in a given disease, how do we organize this information? Thagard writes, “a disease explanation is best thought of as a causal network instantiation, where a causal network describes the interrelations among multiple factors, and instantiation consists of observational or hypothetical assignment of factors to the patient whose disease is being explained.”95 A graphic representation of the CN I for a patient’s developing duodenal 92 Thagard, How Scientists Explain Disease, 113-114. 93 Note that Thagard also is interested in the process of medical discovery and acceptance of explanations in How Scientists Explain Disease. For the sake of my project, I am less interested in this historical development. Although interesting, especially discussions about what social forces work to shape scientists’ decisions to accept a given explanation, this is a large topic that I will not cover in detail. My discussion in this dissertation will be limited to clinical explanations in current debates, and instead focus on questioning the structure of medical explanations. 94 Thagard, How Scientists Explain Disease, 114. 95 Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms”, 61. 62 ulcers (:1 Arthritis 01 painful cor 1 Heavy use (eg. aspir \ CNI ill ugt disease pr Cl exPlanatit diseases. l Smlnure 1 lii‘hind Ct infection, exPlanar 1..» ulcers (as a result of H. pylori infection) can be seen in Figure 1. Environmental factors Arthritis or other Genetic predisposition painful condition (e.g., to increase acid (e. g., smoking, stress) secretion, rapid gastric l emptying, infection) Heavy use of NSAIDs l (e. g., aspirin) Increased acid secretion, Helicobacter pylori infection 1 \ rapid gastric emptying, etc. Gastritis Duodenitis / Duodenal ulcer disease Figure 2 - Thagard’s General CNI for Duodenal Ulcers96 CNI illustrates the causal web involved in disease, both for types of diseases and for the disease process in an individual patient. CN I provides structure for the variety of tasks for which we seek medical explanations. Thagard argues CNI is the proper structure for explanation of specific diseases, e.g., how HIV infection results in AIDS in the human body. CNI is the same structure implemented in other “levels” of medical explanations. CNI is the structure behind explanations of categories of diseases, e. g., how categories of disease like infectious or nutritional diseases function in the body. CNI is also the proper structure for explanations regarding the disease of individuals and of groups.97 Thagard writes, 9‘ Thagard, How Scientists Explain Disease, p 115. NSAIDs are nonsteroidal anti-inflammatory drugs. 97 Again, the important point here is that the CNI explaining a given disease in an individual, in a group, or abstract] y in all humans, is for Thagard structured identically. Differences are only in how we fill in certain levels of detail, or in how we subtract certain causal pathways to reflect what we know about the group or person involved. Even in the discussion here of rather different explanations (e.g., of individuals, groups, or 63 “Explan on) tent abstract medical allows 1 relatitii occurs l tiplana‘ Specific filling it limes [ht Clinical g L Chi ex pl 0i gFOUpt “Explanation of why members of a particular class of people (women, lawyers, and so on) tend to get a particular disease is also causal network instantiation, but at a more abstract level.”98 Thus Thagard believes his CN I is able to do all of the work of the basic medical explanatory schema. Beyond this, CNI provides a richer explanation because it allows for a more detailed causal description, and it is more versatile in that it can be relativized to a number of information contexts. Here, I take Thagard to mean that if we understand the CNI of HIV /AIDS as it occurs in the exemplar of the human body, we have a CNI that is our research explanation. This serves as the basic structure for a clinical explanation of HIV/AIDS in a specific person. We make this move from an abstract human to a specific person by filling in the appropriate facts or making analogies to other known cases. Of course, at times these facts may not be available to us, but the underlying structure of both the clinical and research explanation is the same. Let us consider a more detailed example showing the relationship between the CNI explaining groups’ health and individuals’ health. In explaining the health problems of groups, Thagard writes: People in underdeveloped countries are more likely to have gastritis than North Americans, because poorer sanitation makes it more likely that they will acquire H. pylori infections that produce ulcers. Nuns are more likely to get breast cancer than other women, because women who do not have full-term pregnancies before the age of 30 are more likely to get breast cancers, probably because of some mechanism by which pregnancy affects breast cell division. When we want to explain why a group is more likely to get a disease, we invoke the causal network for the disease factors possessed by members of the group.99 human abstractions), Thagard emphasizes the structural similarities but ignores the pragmatic or contextual features that we must consider in determining how these CNI are fleshed out differently in each situation. 98 Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms”, 61. 99 Thagard, How Scientists Explain Disease, p 116. Thagard retains this language from when he originally wrote this in “Explaining Disease: Correlations, Causes, and Mechanisms”, 75. 64 We knox groups v affecting describi: illlia's c exlllanat light of t Clidence [heads 0 t31‘ng lat exPlanati We know the general level CNI for both ulcers and cancer formation. As we identify the groups we are investigating, we are able to detail more of the relevant causal details affecting this group. Yet CN I also works to explain individual patients’ diseases. In describing why Julia developed an ulcer, Thagard writes: Instantiation of a causal network. . .produces a kind of narrative explanation of why a person becomes sick. We can tell several possible stories about Julia, such as the following: 1. Julia became infected with H. pylori and because of a predisposition to excess acidity she got an ulcer. 2. Julia took a lot of aspirin for her arthritis, which produced so much acidity in her stomach that she got ulcers.l Julia’s clinician, under Thagard’s explanatory model, thus begins with the ideal explanation of ulcer formation. The clinician then “whittles down” that explanation in light of the Julia’s specific case. This is a process of gathering further contextual evidence about Julia’s condition, and evaluating how this information confirms or denies threads of the ideal explanation as relating to Julia’s case. If it is clear Julia has not been taking large amounts of aspirin, say, then explanation 2 above will be rejected, and explanation I is confirmed. Clinical Usefulness of CNI Thagard’s CNI is meant to have practical use and to reflect the structure of medical explanations, both research and clinical explanations. As Thagard writes, “causal network instantiation explanations of the occurrence of both individual and group disease are structural] identical.”l°' These CNI ‘narratives’ ma be similar to those clinicians Y Y '00 Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms”, 74-75. Side note: Thagard repeats this example, but by making it about a generic, unnamed patient, in How Scientists Explain Disease, 1 IS '0’ Thagard, How Scientists Explain Disease, 116. (But see also 115-1 17.) 65 tell to pa approach describe been talc irritation What we Hm. The “finsmitn Benefit EXplan Th exlunatic diniCldflg (A explanato- ailempps {l cl ”Giant tell to patients. As such, clinicians should be able to easily adopt the CNI explanatory approach within their practices. For instance, it seems reasonable that given the CNI described above of ulcer, her doctor might say to a patient, “Frank, the aspirin that you’ve been taking for your arthritis has done damage to the lining of your stomach. This irritation has resulted in an ulcer.” Thagard, though, does not want to overemphasize what we might call the pragmatics of the clinical encounter. He writes: But medical explanation is not just story telling, since a good medical explanation should point out all the interacting factors for which there is causal evidence and for which there is evidence of relevance to the case at hand. A narrative may be a useful device for communicating a causal network instantiation, but it is the ensemble of statistically based causal relations that is more crucial to the explanation. ' Here, Thagard will emphasize the structure of CNI, rather than focus on the process of transmitting this explanation to the patient. I will return to this topic later in this chapter. Benefits and Strengths of CNI Approach to Medical Explanations There are at least four significant strengths to Thagard’s CNI approach to medical explanation. These are that (A) CNI reflects the approach taken by researchers and clinicians towards explanation of disease, (B) CNI utilizes Schaffner’s middle-level theories, (C) CN I provides a useful, formulaic approach to explanation, and (D) CN I has a wide explanatory range. (A) The first benefit of CNI is that it is an evenhanded attempt to reflect the actual explanatory strategies of medical clinicians and researchers. In this way, Thagard attempts to find explanatory strategies that reflect the lived experiences of medical clinicians and researchers, both current and historical. '02 Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms”, 74-75. 66 (E according medicine are not nc‘ allows f0 meehanis to differei CNI abou categories (C cliniciam| exPlanaro be an CXpI medical e, general ca exPlélrtzrtio such that i allows {Or |I (B) CNI utilizing Schaffner’s middle-range theories group disease explanations according to similar causal mechanisms, thus strengthening explanatory unification in medicine. Although CNI relies upon a conception of causal mechanism, its requirements are not nearly as stringent as Salmon’s. While a mechanism is assumed to exist, CN I allows for explanation to occur without having complete knowledge of the causal mechanism.103 Also, by utilizing middle-range theories, the explanatory agents can move to different explanatory ‘levels’ as is needed. They can derive a more detailed oriented CNI about a specific patient, or derive a more abstract explanation about general categories of disease. (C) CNI allows inquisitors to draw from set explanatory patterns. CNI provides clinicians and researchers a working formula about what information to include as explanatory, what information deserves further research, and what information is likely to be an explanatory red herring. The formulaic CNI may allow for generating cleaner medical explanations. For example, this occurs in moving both from the explanation of a general category of disease to the functioning of a specific disease, and also from the explanation of a specific disease to that of a given individual. CNI structures explanation such that it determines what information is explanatorily relevant. Also, the pattern allows for generating new explanations based upon the similarities to other known cases. For example, when AIDS was first recognized, theorists who believed it to be caused by a virus were able to begin with their understandings of how other viruses behaved within the human body. Over time, this knowledge led to the discovery of HIV and an explanation of how HIV is related to AIDS was generated.104 The formula of CNI could '03 While a complete understanding of the mechanism is not necessary for CNI to be useful, I do believe that the CNI is based upon the belief that such a complete or ideal explanation is possible. I continue to argue against such a presupposition. '04 It should be noted that this process is not as linear as may be implied. Research discoveries can lead to generating new or improved explanations, which in turn leads to further discoveries, which in turn can lead to improved explanation. The relationship between explanation and scientific discovery is complicated and deeply intertwined. While Thagard does take up in detailed discussion scientific discovery. 1 do no intend to cover that in any deep way in this dissertation. 67 also be beneficial to clinicians, guiding what information they disclose to patients about their diseases. (D) This ability to explain a wide range of disease topics is also related to the use of middle-level theories, which allows understood analogies and exemplars to be used as the basis for other explanations. On a different point, CNI is rather flexible in that a similar structure can be used to explain categories of disease, a specific disease type, and the occurrence of a particular disease in a particular person or within a given group. This was shown earlier by Thagard’s use of CN I to explain infectious diseases, ulcer formation from H. pylori infection, and Julia’s specific case of developing an ulcer. These strengths of Thagard’s CNI make it initially an appealing model for explanations of disease, both for medical research and for medical practice. Yet I argue in the next section that Thagard’s CNI, while appealing, cannot serve as a proper foundation for generating medical explanations, especially those that seek to answer patients’ why- questions. Thagard’s CNI model makes a common yet problematic move: CNI assumes an ideal explanation is possible, and that it has the greatest explanatory ability. I argue against this conception of ideal medical explanations. Weaknesses and Problems with CNI What are the limitations of Thagard’s CNI? I think Thagard overstates the utility of CN I and other purely causal approaches to disease explanations. The rudimentary explanatory schema Thagard first identifies is of a restricted scope, as I argued above. CNI—Thagard’s explanatory response—is similarly thin. CNI cannot answer many of patients ’ why- questions regarding their health, and it eliminates the patient from much of the explanatory process. Thus, CNI is an insufficient basis for medical explanations. In the following sections I will outline a number of more specific limitations of CNI. First, Thagard’s use of ‘environment’ as a causal category is ambiguous, and leads to a number 68 of difficulties. Second, CNI relies upon ideal explanations to generate clinical explanations, yet such ideal explanations are rarely available in medicine. Third, CN I fails to take seriously the difficulties involved in generating clinical explanations. Fourth, CNI works on what I call a subtractive approach that tries to eliminate information irrelevant to the clinical explanation. Yet this overlooks a number of ways in which adding additional information is required for clinical explanations. Definition of Environment The first problem I identify with CNI is different from the rest. The concern is about the role of ‘environment’ as a source of causal information. The critique holds for both research explanations and clinical explanations. The additional critiques focus primarily upon limitations of CNI for clinical explanations. One of my concerns with standard theories of medical explanations, including Thagard’s, is they fail to account for social factors as explanatory of diseasems Thagard’s CNI can technically handle this, but only by clumping “social factors” under “environment”. The term ‘environment’ remains so ambiguous under CN I that it can account for nearly any type of phenomenon. For example, we can easily accept that toxic chemicals at the workplace may be explanatory of employees’ health; yet what should we say of a stressful workplace? Is it, too, toxic? ‘Environment’ as a catch-all category loses its strength in categorizing and explaining. The first weakness of Thagard’s CNI involves the lack of definition in describing ‘environment’ as a causal factor of disease. Since Thagard explicitly argues for understanding disease as multi-factorial (or against understandings of disease as having single causes), we can ask what role does environment play in disease? Here, I point out a number of ways Thagard may be '05 Recall that Uncle Bill finds racism (that is, the racist attitude of his boss) to be explanatory of his health. In the case of J.S., gender is intimately tied to her “unexplained” medical condition. Even “stress”—a nearly ubiquitous factor in disease causation—is involved in Julia’s case, but it can have radically different social meanings when described in greater detail. 69 conceiving of environment, yet each of these implies something different for properly shaping CNI. Environment is everything: In describing environment, one strategy is to define it as all that is left over from the causal factors of primary interest. For example, this conception of environment would allow that Julia’s ulcer is explained in some way by the physical space of her home, the air quality and altitude of the city she lives in, and the color car she drives. In this way, ‘environment’ can become so all-inclusive that every nuance of Julia’s day can be considered an environmental factor contributing to her ulcer. Yet this conception of environment is so open-ended as to be unhelpful. It cannot differentiate between useful aspects of the environment (say, environmental toxins) from useless details (say, car color). While it may be possible to draw some sort of causal connection between Julia’s ulcer and her car color, it is unlikely that this would be helpful as part of a clinical explanation. Environment as consistent background: It may be instead that Thagard has in mind a conception of environment is somehow a background that is held consistent. But this, too, is difficult in that Julia’s city life may be dramatically different from that of a rural farmer. One solution may be that environment is closer to a set of Mackie’s INUS conditions: an Insufficient but Necessary part of a condition which is itself Unnecessary but Sufficient for the result.106 It is true that bacterial infection could have caused Julia’s ulcer, but the bacteria are not necessary for ulcer formation. Consider that some medications have been determined to be the cause of approximately 20% of ulcers. '07 Bacterial infection and a few other conditions may result in ulcer formation, but bacterial infection is not necessary for ulcer formation. If this is the case, Thagard does not clearly '06 J. L. Mackie, “Causes and Conditionals,” American Philosophical Quarterly 2 (1965): 245-65. “’7 Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms”, 70. 7O distinguish when we should consider, say, viruses or bacteria as ‘background’ or ‘environment’, too. By not addressing this problem clearly, we run into specific problems in both research and clinical explanations. What conditions might the explainer assume are held constant (and are those assumptions justified)? What basis do we have for assuming these conditions to be constant, but not others? In clinical explanations, how do you explain clearly the complexity (and possible uncertainty) of the causal import of environmental differences? More importantly, it is likely that clinicians and patients will make such determinations differently, given their differing backgrounds and scientific abilities. Environment and domination: There are important political reasons to question our motives for emphasizing some facets of the body as ‘active’ and others as ‘background’; feminist philosophers of science have best raised this discussion. For one such example, consider E.F. Keller’s study of DNA, which is often referred to in biology and medical textbooks as the “master molecule”. 108 Keller rightly questions our scientific and cultural fascination with this molecule. The master molecule conception of genetics holds that DNA influences changes upon the cell, is the source for information, but is itself stable and unchanging. DNA directs the chaotic activities of the surrounding cell. Here, our descriptions of cellular activities are dramatically shaded by conceptions of hierarchy, domination, and control. Repeating these patterns either unreflectively or unnecessarily as a part of genetic and non- genetic explanations is a problem that Thagard does not remedy in his theory of medical explanations.‘09 '08 Evelyn Fox Keller, Reflections on Gender and Science. (New Haven: Yale University Pres, 1985) 150. '09 Notice that although these biological and medical texts and explanatory narratives of hierarchy, domination, and control are common, they are changeable. For a good example of this change, see Emily Martin’s retelling of the fertilization story, where the egg (not the sperm) is the active participant in “The Egg and the Sperm: How Science has Constructed a Romance Based on Stereotypical Male-Female Roles.” This example also raises a point that I will address again later: that the line between epistemic and political projects is not a sharp distinction. See Lorraine Code, Epistemic Responsibility (Hanover, CT: University Press of New England, 1987). 71 Environment and social factors of disease: Here, I return to the original worry I described at the outset of this section. Consider the question of whether ‘environmental factors of disease’ includes or is separable from ‘social factors contributing to disease’?1 1° Let us consider Julia’s case, and fill in a bit more of her background. Is it explanatory of her ulcer that she is in a “high stress” career? I think Thagard would likely say yes. But let us fill in some hypothetical details. Although Julia would love to quit, her life circumstances prevent her from finding another job. She is the only woman in her firm, and feels that her work is constantly scrutinized as others wait for her to make a mistake. Julia has also been “outed” against her will, and she endures daily homophobic harassment at her law firm. The homophobic and sexist conditions Julia endures may be real, but are they explanatory of her ulcer? Here, I am unclear what Thagard would say since he never addresses this issue. Again, he opens up an interesting case, but doesn’t push the nuances of it. Under both renditions, to cite ‘stress’ as the relevant causal factor seems accurate yet incomplete. The source of the stress is not part of the natural environment in the same way as temperature or air pressure. Julia’s stress is due to social relationships. Let us first consider another example. What of Uncle Bill’s story, which was discussed in the Introduction? Recall that Uncle Bill claims his health problems were caused by the racist attitudes of his employer, which contributed to an unhealthy—literally, a toxic—workplace. For the sake of this example, let us grant that Uncle Bill’s employer truly is racist and unworried about his employees’ health. Thagard’s CNI would easily allow for workplace toxins to be explanatory of Uncle Bill’s stroke. But is the fact of the employer’s racist attitudes explanatory of Uncle Bill’s illness? “° Although Thagard does acknowledge some social factors, these are mainly historical factors that contribute to clinicians accepting of explanations, i.e., (dis)belief that bacteria are involved in gastric ulcers. Thagard does not acknowledge that social factors play a role in the explanations generated by individual clinicians, say, during clinical visits. 72 One strategy for Thagard may be to explicitly question how far back in a causal chain must an explanation stretch. It seems clear that ‘stress’ and ‘toxic environment’ are salient, proximal medical features of Julia’s and Uncle Bill’s stories. But if we look at the more distal causes, we would uncover these damaging social forces. Although both may be right, Thagard gives no argument for why the more proximal causes ought to be prioritized. Thagard does not explicitly argue for this, but it is implicit in his examples and his lack of addressing directly the social determinants of health. Yet his lack of considering social forces is in and of itself telling. It may be that Thagard assumes social forces influencing health are environmental, just more distant. It may be that Thagard sees these social forces as possibly being real, but not relevant to medical explanations, valuing instead more proximal and physical determinants of health. It is worth noting that Thagard does consider some social forces as part of generating medical explanations. He limits these, though, to the social forces that are part of inquiry and theory acceptance within scientific research communities, such as the social processes of peer-review of data and developing medical consensus for an explanation of a given disease.l '1 This is too thin a reflection upon the social forces at work in medicine, since it identifies the social relationships and experiences of clinicians and researchers as explanatorily relevant, yet it erases the experiences and participation of patients, making them explanatorily irrelevant. Given the structure of CNI, and the fact that enduring homophobic or racist assaults and ridicule lead to stress, these and other social determinants of health are explanatory yet excluded from consideration under CNI. Their influence on patients’ health and morbidity is scientifically measurable.l '2 The people subjected to these ”1 Again, I believe this to be a symptom of the difference between Thagard’s and my original interests in medical explanations. I have begun from the position that medical explanations ought to improve patients’ understanding of health (clinical explanations). Thagard has begun by understanding medical explanations as part of biomedical research (research explanations) and the historical discovery of explanations. For more on his concept of social interactions in biomedical research, see his Chapters 6, 11, and 12. ”2 M. G. Marmot, George Davey Smith, et al, “Health Inequalities Among British Civil Servants: The Whitehall II Study.” The Lancet. 337. 8754 (1991): 1387. This study repeats and confirms the original 73 stresses often locate their health problems not in a vague concept of “stress”, but the source of that stress, just as Uncle Bill has done. But the common move of Thagard and other clinicians to group social experiences as ‘environment’ erases them from consideration by hiding their importance to good medical explanations as salient causal factors. If we were to develop a medical explanation theory that takes such social determinants of health seriously, it would raise some interesting problems. First, it would add another level of complexity to what is already a confusing situation. This need not turn us away from considering them; instead, it shows more work that needs to be clarified. Second, the pragmatics of clinical explanations are more complex than were originally thought. Clinicians know in some way that social situation has a strong connection to health, but those same clinicians may be unsure of how to provide proper explanations on the topic to their patients. Beyond knowing that homophobia or racism is explanatory of Julia’s and Uncle Bill’s health, a doctor would need to consider the when’s and why’s of bringing this up to patients. For instance, it may be awkward or condescending for a white, privileged male doctor to tell Uncle Bill that racism exists, and that this may not be a virtuous part of his life. In another case, it may involve telling someone of a low socio-economic status that poverty is not good for his or her health.1 '3 So, politically savvy and well-intentioned doctors may want to have such conversations with their patients. Yet there are negatives in pointing this information out to the patient: it may be obvious, and likely condescending. Although the structure of the medical explanation may include such social determinants of health, it is another matter to properly navigate how (or whether) to bring this information up with a patient as part of a clinical explanation. This is not to say the explanation should be avoided, but it findings in a previous study that socioeconomic status is inversely proportionate to morbidity. I thank Jim Nelson for pointing me to this example. ”3 Janny Scott, “Life at the Top in America Isn’t Just Better, It’s Longer”, New York Times Online, http://nytimes.com, accessed May 16, 2005. 74 points out the difficulties of having such explanatory conversations with patients. This may be yet another factor which disinclines clinicians to use such information in generating clinical explanations. Yet a good theory of medical explanation should help us identify and resolve such miscommunications. CNI & the Myth of Ideal Explanations Recall that the first explanatory schema Thagard describes was too basic. Although there was flexibility in filling in the proper terms, this did not provide guidance as to how much detail (or under what circumstances) a concept needed greater clarification. Under Thagard’s CNI, the complete causal network—generated at the middle-range—describes all of the causally relevant features. Thagard’s CNI develops out of a long tradition in theorizing explanations in the philosophy of science. As such, explanations are viewed as a list of causal factors, organized so as to be meaningful. Such accounts are often highly detailed and include a wide range of biological theories, environmental, chemical, and genetic data. When laid out in full detail, CN I explanations are quite similar to Railton’s ideal explanations and Salmon’s mechanistic explanations. Well-crafted CNI include all of the causally relevant details, as do Railton’s ideal explanations. Thus, research explanations—here, think of Thagard’s CNI for formation of ulcers or cancer tumors—identify all of the possible causal contributors. As such, CNI can become complicated, extensive explanations.‘ '4 Yet this conception of explanation implies that explanations are set features of the world that are discovered by research, rather than tools that are created for a specific project (a project determined by the local circumstances) as fits the erotetic view of m Recall Thagard’s quote, discussed earlier, that “a good medical explanation should point out all the interacting factors for which there is causal evidence and for which there is evidence of relevance to the case at hand. A narrative may be a useful device for communicating a causal network instantiation, but it is the ensemble of statistically based causal relations that is more crucial to the explanation” (Thagard, “Explaining Disease: Correlations, Causes, and Mechanisms, 74-75). Side note: Thagard repeats this example, but by making it about a generic, unnamed patient (How Scientists Explain Disease, 115). 75 explanation. Thagard’s mistake here is that he takes part in what I call “the myth of ideal explanations.” As part of this myth, explanations are seen as independent features of the world. But rather than reflecting a ‘view from nowhere’, CN I reflects a rather specific point of view.1 ‘5 CNI contain the information that is determined as relevant by clinicians and researchers. CNI, like all explanations, are tools generated by individuals or groups for a specific purpose. Research explanations in particular tend to take on the structure of ideal explanations. Such explanations are generated over long periods of time. They are meant for use by researchers and clinicians (i.e., scientifically trained folks). The highly complicated nature of such explanations is not seen as a liability, but as a reflection of the complicated features involved in the given disease. If research explanations are CNI’s equivalent to ideal explanations, then clinical explanations are non-ideal explanations (or what Railton calls ‘explanatory information’). Clinical explanations are parts of the ideal explanation. How does this transition from ideal to non-ideal explanation happen? Again, Thagard does not give clear guidance, but I believe the following example is a generous read of what that process might be like. Julia reports as part of her medical history that she has not been taking nonsteroidal anti- inflammatory drugs (NSAIDs), e.g., aspirin. This eliminates NSAIDs, which is the cause of about 20% of ulcers. Julia’s doctor can then focus tests on determining whether she is suffering from H. pylori infection. The move from ideal to non-ideal explanation invokes the subtractive move of eliminating non-relevant aspects of the ideal. To the extent this is possible, focus remains on relevant threads of the explanation. These explanatory threads are then expanded for their level of detail. Because Thagard conceives of CNI in a way that is similar to Railton’s ideal explanations, this theory encounters certain problems. The first limitation I explore is that "5 This concept of ‘the view from nowhere’ or ‘the view from everywhere’ has been criticized widely, especially by feminist philosophers of science. 76 CN I explanations are often too complicated, much like ideal explanations, and thus these often do not answer what patients want to know. As such, these cannot serve as the basis for medical explanations in clinical practice. Explanations are not only lists of causal factors because lists of causal factors alone are not explanatory. Here, as I argued in the Chapter 2, I agree with van Fraassen: knowledge of causal chains by itself is not explanatory because explanations are necessarily contextual. A context is needed from which to ask a why-question, and then to answer the question by referencing the causal information. Explanations require editing and responding to context of the questioner.1 ‘6 In this way, clinicians serve as mediators (or midwives) who deliver ideal explanations from a scientific domain and present them to the patient in a meaningful way. Yet neither Thagard’s CNI nor Salmon’s or Railton’s responses provide adequate guidance on how to navigate the pragmatics of generating non-ideal explanations from ideal explanations. They argue that relevance determines which information is part of the explanation, and which information is non-explanatory. The worry about explanatory uptake is not a concern of explanatory structure. I will next argue that this move is not nearly as straightforward as these authors believe. CNI’s Failure to Consider Seriously the Difficulties of Generating Clinical Explanations At some points, I have criticized Thagard’s CNI as being too thin a response to the problem of medical explanations. In large part this is because Thagard too narrowly conceives of the work that medical explanations do. Thagard sees the move from ideal research CNI (explanations of a disease type) to explaining disease in groups or individuals as involving only the process of detailing appropriate causal information. Yet this is to overlook a wide range of difficulties involved as part of generating clinical CXPI anations. l I 6 I take up this project in more detail in Chapter 6. 77 Eac to d SCI: exp pan gen eon the: gen be : can 10p Sin afln ICC: acl [her 8091 Drag When patients see doctors, their cases will not be “textbook perfect.” Because of this, the translation from ideal explanations to clinical explanations will not be smooth. Each patient’s condition may be different. So clinicians must develop over time the skills to diagnose and identify relevant causal factors, and explain these to patients. Finally, this set of skills will take time to develop. Navigating Levels and Details Thagard’s middle-range CN I is structurally similar to Railton’s ideal explanations. CNI are narratives of all the possibly relevant causal information about a particular disease type. Even if we were to grant that such ideal explanations can be generated, this is not a useful tool (by itself) to provide to clinicians. If ideal explanations contain all possibly relevant information, the clinician is left without guidance as to how they should extract the proper portions of this ideal for use with patients—that is, how to generate a non-ideal explanation from an ideal explanation. For instance, questions may be about what “level” the explanation should be given, or whether the request is for causal information or for treatment options. Thagard’s CNI gives no guidance on this topic. Similarly, CNI fails to address how detailed a clinical explanation ought to be. Since CNI are based on ideal explanations, they will often contain an overwhelming amount of detail. It is likely that this will be too much for many patients to absorb. But recall Sandra Harding’s important point from Chapter 1: if explanation fails to be grasped by the human mind, then the explanatory project has failed. If a patient cannot understand a clinical explanation—if only clinicians/researchers can understand the explanation— then it has failed. Navigating Pragmatic and Contextual Considerations There are also pragmatic considerations about the explanatory experience—the SOCial and psychological nuances of clinical explanations—that are separable from the praglmatics shaping the underlying structure of the explanation. Patients’ topics of 78 interest, their educational background and familiarity with medicine, their plans for how they may respond to the disease, and other contextual factors are set by the interests of the patient. As such, there is a wide array of pragmatic and contextual features that the clinician ought to consider while developing a clinical explanation that the patient can utilize. There are extra-explanatory considerations that come into play in developing a clinical explanation. In this way, medical explanations are significantly more complicated than Thagard’s and Schaffner’s approaches have implied. This also means they are philosophically more interesting than had previously been granted. Moral, legal, and other ‘practical’ considerations must be taken into account as part of a clinical explanation. Thus, the explanatory schemas for clinical explanations will be different from those that work for other scientific projects, and perhaps even different from biomedical research explanations. For example, clinical explanations are immediately complicated by the background knowledge and explanatory needs of the patient. Although this will be examined in greater detail in the following chapter, for now we can discuss a few of these basics. For example, what role does emotion have in explanation? Trust, for example, plays a significant role in explanation uptake. Patients often grant one form of trust because clinicians are seen as experts.1 '7 Another form of trust comes from long-term relationships, doctor-patient conversations, and clinicians admitting their scope and limitations of their knowledge. Surely other emotions and social virtues have a role in explanatory uptake: intelligence, compassion, lack of condescension or arrogance, patience, and humor. CNI gives no guidance about the additional pragmatic features of l I 7 John Hardwig, “Toward an Ethics of Expertise” in Daniel Wueste, ed., Professional Ethics and Social Responsibility (Maryland: Rowman & Littlefield, 1994), 83-101. He addresses the problem of relying upon exPerts of all sorts: how ought laypersons judge the experts’ claims to knowledge? For a different perspective, see also Lorraine Code’s Epistemic Responsibility. 79 the explanatory conversation that would ensure patient uptake of the explanation. Without such uptake, the why-question has not been answered. Here, one might object that by worrying about the pragmatic considerations of psychological—not merely the structural—components of medical explanation, I have opened up a can of worms with no reasonable hope of training them to be orderly. Consideration of all of the pragmatic factors—which will likely change from person to person, and perhaps daily for each person—is such an overwhelming task that it is unmanageable. There is something seductive about Thagard’s failing to take up this task, since the answer will likely be overly complicated.l '8 By avoiding this morass, Thagard’s explanatory solution is much neater; he need only identify the relevant causal factors. I myself will not consider all of these possibilities. I will limit my general discussion to the pragmatics of explanation, but I will not address in a strong sense the role of emotion. Although CNI provides a simple, formulaic response, it is undesirable in that it misses much of the complexity of medical explanations. It again requires a false sense of the work done by medical explanations and the real-world contexts in which they are generated. If Thagard wants CN I to benefit clinical practice, then a theory of medical explanation ought to account for the messiness of the clinic. The Subtractive Method for Clinical Explanations There is an additional role for the pragmatics of explanation that would be articulated before an explanatory theory like CNI could be useful to clinical explanations. In describing the move from ideal explanation to non-ideal explanation, I argued above that the clinician begins with the ideal CNI for a given disease. The clinician eliminates threads of this explanation that are not relevant to this patient’s case. In this way, clinical explanations are parts of the (ideal) research explanation. Still, it may not always be clear us salmon makes this point in Four Decades of Explanation: articulating the pragmatics of explanation (gray be an overwhelmingly complicated project. Thus, projects involving the articulation of a pragmatic cox—y of explanation are either doomed to failure or run the risk of remaining frustratingly incomplete. 8O 61: CI cl. Cl C! B: $11 The flip what guidelines the clinician should use for this subtractive method, except to try to eliminate the non-relevant explanatory threads. The problem, though, with the “subtractive” method of generating clinical explanations, is that it gives a false sense of how explanations are generated in real clinical encounters. The clinical explanation—even as a whittled-down non-ideal explanation—can often be highly complicated, full of scientific jargon and complex causal mechanisms. The editing process from ideal explanation to non-ideal explanation, when conceived of in this formal way, does not value patients’ uptake of the explanations; the value is placed only on eliminating irrelevant features. While this, of course, has value, it is not sufficient to make clinical explanations useful to patients. Below I consider three problems raised by the subtractive method. First, it ignores alternative explanatory strategies, including the use of conversation aids such as metaphors, or the possibility of the pragmatics of explanation generally. Second, additional explanations may be required, rather than whittling down the primary explanation. This is what I call the “additive method”. Finally, the goals of some clinical explanations may be to keep explanatory “gaps” open, rather than to try to reduce them. First, consider that many clinical explanations may employ metaphors, analogies, and other heuristics to translate the scientific into lay terms. Such work is ignored by the subtractive method. Second, there are times when an additive method may be called for. Consider that Julia’s mother had been treated for ulcers in the past. Julia is scared and frustrated that this condition “runs in the family.” Julia’s mother, though, took large quantities of aspirin for her arthritis; this ultimately resulted in ulcer formation. Julia’s ulcer, though, has not been caused by damage from drug intake, but instead from a hostile work environment WhiCh made Julia more susceptible to bacterial infection. While discontinuing pain medication may have helped Julia’s mom, it will not help Julia. A proper clinical 8x131 anation for Julia will need to explain why her mother’s case is different from her 81 own. In this way, this clinical explanation requires an additive method—an additional explanation of the mother’s case, and how that is different from Julia’s own case. To see clinical explanations as only whittled-down versions of ideal research explanations misses important types of explanatory information. Finally, in other cases, the “subtractive” method too quickly works to close explanatory gaps. In some clinical explanations, the goal is not to close explanatory gaps to gain understanding, but instead to keep these gaps open. Consider the case of J .S.: both she and her clinicians have been seeking a diagnosis, yet no single diagnosis seems to fit. There are a number of pressures to give her condition a name (e. g., to better develop a treatment, or to be reimbursed for care by insurance agencies). But each diagnosis works against J.S.’s understanding of her medical condition, rather than improving it. Some may object here, arguing that explanations are meant to improve understanding by showing how the world works and does not work. For clinicians especially, treatment decisions will demand a specific understanding of the case at hand in order to determine the best possible therapeutic response. But notice that such a explanatory and therapeutic response has not helped J .S. in the past. In her case, it may be appropriate to remain undiagnosed, or to continue to seek alternative (even seemingly contradictory) explanations, rather than reduce her condition to a poorly fitting diagnosis. Towards Strategies for Correcting Clinical Explanations Below, I offer two possible corrective means to refocus how we understand the role of clinical explanations and how they are generated. These themes will then be articulated in detail in the following chapters. Designing Explanations & Keeping Explanatory Space Open If the subtractive method of generating clinical explanations is faulty, I argue a Possible solution lies in what I am calling examining the activities of ‘designing 82 Cl Ca de de 801 eth ah sitr COI ell 8x; CXp War explanations’ and ‘keeping explanatory space open.’ Here, I draw from the work of Caroline Whitbeck and Margaret Urban Walker. Caroline Whitbeck argues that applied ethics problems are similar to problems of design faced by engineers. There will be no singular correct solution (to problems of design or of applied ethics). Instead, how we set up the problem, what interests we prioritize over others, will determine better or worse solutions or responses. 1 ‘9 Inevitably, some response strategies will do better than others. Our responses require a certain amount of awareness of the problem that generated the request (for explanation or for ethical response), creativity to understand the problem (the questioning context from which the patient is asking for help), understanding of the complexity and nuance of the situation, and to find a strong solution. If this is correct, then crafting an explanation should be understood less as work of uncovering (discovering) relevant causal phenomena, and more about crafting/designing an explanation that meets the needs of patient/questioner, and the context of the problem. (I think this is both about understanding the structure of explanations, and the goals of explanations: both answer what is a ‘good’ or ‘better’ explanation.) This view of generating explanations ties nicely to the metaphor of explanations as a tool to improve understanding that I developed in Chapter 1. In many ways, this is in line with Railton’s distinction between ‘ideal explanatory texts’ and ‘explanatory information’. What bioethicists can contribute here is a better navigation of general differences between clinicians’ and patients’ goals for explanations. It will be unwieldy (and undesirable) to generate ideal explanatory texts in response to patients’ questions, and ideal explanations often will not answer patients’ why-questions. So, all clinical explanations will veer away from this ideal, and instead be a variation of CXPI anatory information. But determining which interests shape the explanatory 119 Cat‘oline Whitbeck, “Ethics as Design: Doing Justice to Moral Problems” Hastings Center Report 26 (I 996) : 9-16. 83 ‘____ inic lll" exp gen clin desr ope: out and to rr see ' Wor then gene ager exp] C0r15 expl lulle more: rIllnu relati information provided is not obvious. Yet clinicians can do more than determine which types of causal information are most salient. Clinicians should be interested in explanations to see whether patients’ needs are being met by medical explanations being generated. This explanatory work, though, typically occurs within the physical space of the clinic as part of conversations between embodied persons. Margaret Urban Walker describes a primary role of ethics committees in the clinic as ‘keeping moral space open’.‘20 By this she means that ethics committees must worry not only about the moral outcomes of ethics consults, but also about the physicality and social dynamics of where and when ethics is done. For instance, having time scheduled and office space allocated to moral deliberations allows clinicians permission to explore these issues, sometimes to see this reflection as expected of them. Also, this gives bioethicists the time to do their work in guiding such discussions. Adapting Walker’s analogy to my discussion of clinical explanations, I argue there are features of ‘explanatory space’ that are under-theorized. For example, the generation of medical explanations requires time and a location for the proper epistemic agents to come together and discuss the problem and then begin to formulate an explanation. Most of these conversations will take place in a clinician’s office. But time constraints will often be a negative force influencing the generation of clinical explanations. A relationship between the clinician and the patients may be necessary for fuller understanding of the patients’ needs to be understood. This may involve ongoing notes/charts to be kept and consulted. Yet clinical encounters are ever shortened (10 to 15 minutes), and patients’ changing insurance agencies prevent them from creating relationships with a single clinician. This often results in doctor-patient relationships Where patients’ interests may be unknown to the clinician. Under such conditions there is 120 33 Margaret Urban Walker, “Keeping Moral Space Open”, Hastings Center Report 23 (March-April 1993): 84 the danger that doctors will revert to developing clinical explanations shaped by clinical and scientific features, rather than those that may be more meaningful to the patient. At this level, bioethicists need not engage in debates about the theoretical aspects of an explanation. Instead, there is additional work to be done in determining how physical features of the clinic and pressures of the health care system influence the generation of explanations. Time constraints prevent detailed history taking that may be required for patients to provide information to their doctors. As this communication is shortened, there is less time for the clinician to gain insight into the social dynamics affecting the patient. Shortened clinical visits may result in unhappy patients who feel unheard by their doctors, just as Uncle Bill and J .8. report. Their stories are complex and take time to tell. When time constraints prevent this, the clinician is prevented from having necessary explanatory clues and information required for generating clinical explanations that will help the patient to understand better. In large part, philosophers of science, including van Fraassen, have ignored such practical, “on the ground” concerns of generating explanations. Their focus has instead been on the structure of explanations, and these concerns described above are external (and therefore uninteresting) to the structure of explanations. There are additional pragmatic concerns that van Fraassen did not account for in his original conception of pragmatic explanations. Despite van Fraassen’s failure to list them, these practical concerns are important, and this comes to light especially when describing medical explanations. Medical explanations are not just epistemic projects (as might be explanations of physics), but ones with the goals of improving patients’ lives through enriching their ability to make health decisions. In these situations, I see the role of bioethicists not as grappling with explanation theory. Instead, they have a role in e Valuating explanations (seeing whether they meet patients’ needs), and what aspects of the Clinic contribute to the quality of explanation generated. 85 Determining the Epistemic Agents of Medical Explanations Finally, in addition to the above conversations of how explanations are structured, generated, and evaluated, I worry about who can generate medical explanations. This is not a critique unique to Thagard’s CNI, but instead a common problem arising throughout the literature on medical explanations.‘21 Although Thagard remains largely silent as to who are the agents of medical explanations, it is implicit in his examples that only clinicians and researchers generate CNI. In one sense, this may not be surprising. We go to the doctor’s office for advice when we are sick, and we can reasonably expect from them explanations of what is wrong. As medical experts, doctors ought to undertake the task of generating explanations. While common, this is neither the only nor the best approach for generating medical explanations. One might object here that perhaps the defect is in Thagard’s choice of examples, rather than with a theoretical flaw in CN I. In response, I would repeat that Thagard does not address the explanatory needs of patients, nor does he raise this as a possible concern. But again, maybe that was an unintentional oversight. If Thagard did seek to include patients in the generating of CNI, could he? I would argue he cannot, without making significant changes to CNI. Most importantly, under CNI only clinicians and medical researchers have the epistemic authority to make knowledge claims about CNI. While Thagard worries that both clinicians and “ordinary folks” should be able to use CN I, only clinicians can generate medical explanations due to their authority and extra-ordinariness 0-0» medical and scientific training).122 121 For example, I also take Kenneth Schaffner’s Discovery and Explanation in Biology and Medicine to be symptomatic of this problem. His analysis focuses primarily upon biomedical researchers as the agents of medical explanations, but he does not address in any way who these agents are. Nor does he question whether it is appropriate or desirable to consider an expanded conception of epistemic agents. 122 This is Thagard’s term. See his “Explaining Disease: Correlations, Causes, and Mechanisms”, 68 and 70. Thagard continues this language in How Scientists Explain Disease. While his word choice may be unfortunate and unintentional, I believe it points to a certain prioritizing of clinicians over patients implicit in Thagard’s basic conception of medical explanation, which leads to certain problems I address below. 86 surely Resear inlorrr create: comp: such [ Under an improved theory of medical explanation, this need not be the case, but surely it is an accurate description of most current medical explanatory projects. Researchers and clinicians are generally the only ones allowed to determine what information is explanatory. Why is this the case? This domain of medical authority is created and policed. Doctors comprise a privileged and powerful group (especially when compared to patients). As Hilde Lindemann Nelson explains, because clinicians have such power, they are able to employ certain tactics: for maintaining cognitive authority and control [by insisting] that only certain kinds of knowledge “count”—namely, the kinds that the authoritative knowers themselves have authorized—physicians are continually tempted to affirm the very practices that perpetuate their inability to see [from the point of view of patients]. That is, they are continually tempted to discount what their patients know, which then makes their own claims to knowledge appear inevitable and right. As Margaret Urban Walker points out, reducing, circumscribing, or discrediting the status of those further down the epistemic hierarchy constitutes a kind of “epistemic firewall” that insulates those in authority by allowing them to dismiss the knowledge claims of those below and “prove” their unreliability as judges.123 In this way, if patients are excluded as explanatory agents, then explanation theory has become a tool not for epistemic enlightenment but for gaining and maintaining control over patients. Thagard’s CNI never requires clinicians to fully listen to or see their patients as full persons who are capable of contributing to the generation of explanations. While clinicians have been criticized generally for not allowing and facilitating patients to make informed, autonomous decisions, the problem typically address only moral decision making, overlooking patients’ epistemic roles. But an analogous situation exists in generating and evaluating medical explanations; the role of patients as active, participating epistemic agents warrants further analysis. Yet under current theories of medical explanation, including Thagard’s CNI, it is ensured that only clinicians and '23 Hilde Lindemann Nelson, “Knowledge at the Bedside: A Feminist View of What’s Happening with This Patient”, The Journal of Clinical Ethics 7. l (1996) 25. 87 researchers are allowed to be epistemic agents of medical explanations; patients are relegated to the role of subjects of explanation. I argue this requires remedy. Thus far, I have argued that CNI allows clinicians to generate—and when necessary, edit—medical explanations. This explanatory strategy limits the role of the patient to that of epistemic subject; the idea of the patient as an active, participating epistemic agent has no place in Thagard’s CNI.124 I will not solve this problem here; I only raise it as a problem. I return to this problem in greater detail in Chapter 4, where I provide clearer grounds for understanding patients as active epistemic agents who both generate explanatory requests and explanatory responses. Conclusion What is the proper theoretical structure for medical explanations? As a means to answering this important philosophical question, I began this chapter by describing the structure of the medical explanation theory argued for by Paul Thagard: medical explanations as Causal Network Instantiations. I take CNI to be one of the best articulated theories of medical explanation, and it begins with the important goals of trying to meet the explanatory needs of both biomedical researchers and of patients. While CNI is often illuminating, I have argued it is not a sufficient basis for clinical explanations of disease. While this schema has certain benefits, it remains problematic as a basis for generating medical explanations. CNI shares a number of structural similarities to Railton’s ideal explanations, which makes CNI problematic to construct good medical explanations, especially as part of doctor-patient conversations regarding patients’ illnesses. 124 At most, the patient is “active” in the sense of being a source for relevant data, e.g., “I have had a fever of 103° for 20 hours, and a scratchy throat. I’ve not been out of the country in the last six months.” But this is not the same as saying the patient is active in shaping or generating the explanation; although reference manuals are useful, they are not explanatorily active. 88 Thagard fails to take into consideration a wide range of work that medical explanations are called to do, and the variety of situations in which research and clinical explanations are generated. Thus, CN I is too thin an account of medical explanations. An improved schema of medical explanations should account for social factors as explanatory of disease, that is, it should explain how racism, sexism, and homophobia, privilege, wealth, and poverty are explanatory of disease. In addition, clinical explanations—in order to be truly explanatory—require patients’ uptake. Ensuring such uptake is a complicated process, one that involves the use of pragmatic considerations. Yet Thagard’s CNI masks such complications. Finally, CNI wrongly relegates patients to the position of epistemic subjects, thus eliminating the possibility for patients to serve as epistemic agents with the ability of participating in generating medical explanations. These are the limits of current theories of medical explanation, which leads me to ask in the next two chapters, respectively: Who are the proper agents of medical explanation? And what is the proper role of pragmatic considerations in shaping medical (clinical) explanations? The latter is trickier, more complex than Thagard thought, but not so overwhelming as to be unmanageable, as is implied by Salmon and Railton. To provide a structure to the consideration of pragmatic factors in generating medical explanations, I will draw from the work of Bas van Fraassen on the pragmatics of explanation. I will argue for an explanatory schema that allows patients and clinicians to collaborate both in asking why-questions about patients’ health, and in generating explanatory responses to these why-questions. Before discussing the pragmatics of medical explanation, I will argue that patients must be considered as explanatory agents in medical explanation. Drawing on the work of Lynn Hankinson Nelson, I will argue for understanding the agents of explanations as communities: of clinicians, of patients, and of doctor-patient collaborations. After showing how clinicians and patients can be the agents seeking medical explanations, I go 89 on to argue they must navigate the contextual features of a pragmatic medical explanations. CHAPTER 4: WHO EXPLAINS DISEASE?: EPISTEMOLOGICAL COMMUNITIES AND MEDICAL EXPLANATION 8 Introduction In the previous chapter (Chapter 3) I argued that Thagard’s explanatory schema (medical explanations as Causal Network Instantiations) is problematic as a basis for medical explanations. First, CNI makes the patient a passive recipient of medical explanations, and the clinician is the active agent. Second, CNI does not provide the theoretical resources for understanding patients’ explanatory requests and for ensuring that patients understand the explanations they are given. Third, CNI assumes clinical explanations are parts of biomedical-research explanations, rather than viewing clinical explanations as generated on their own terms (and for their own set of goals and goods). One might argue that essentially all models of explanation that have been applied to medical explanations have made these problems. I have focused my critique on Thagard’s CNI in that it is one of the best articulated theories, and I take it to be representative of current standard theories of medical explanation. These limitations are most obvious when considering the construction of clinical explanations, those between doctors and patients, and how such explanations fail to answer patients’ why-questions about their health. In this chapter, I articulate the role of the patient and the relevant power imbalances that exist between clinicians and patients, which influence the generation and evaluation of medical explanations. Bioethics consistently worries about power imbalances in medical practice, say, as part of health care practice (e. g., analyses of power in doctor-patient relationships and doctor-nurse relationships, avoiding 91 paternalism, or patients’ reproductive freedom) or in the activities of biomedical research (e. g., analyses of power imbalances in the researcher-subject relationship lead to requirements for patients’ informed consent before entering clinical trials).125 While bioethics has questioned the role of power in clinical practice and in generating medical knowledge, the role of power in the context of generating medical explanations has remained thus far unexamined. Thagard’s medical explanation theory does not allow room for an analysis of how power influences the generation of clinical explanations, nor does CNI recognize that clinical explanations are generated by embodied persons whose social context may influence the explanation. This exclusion poses a set of interesting problems. First, the role of the patient in generating medical explanations is erased. Second, this strategy recognizes only clinicians—as a stand-in for the scientific expert and the holder of medical knowledge—as the agents of medical explanations. Thus, the patient must seek the explanations from a clinician qua expert. A good CNI explanation is generated by the clinician, who needs not take into account the explanatory needs of the patient. CNI does not require the clinician to generate the explanation for the sake of the patient’s understanding. Clinicians may do so out of a sense of beneficence (that is, it is good for patients to understand medical explanations), or perhaps it positively impacts patients’ health. However, there is neither a sense that CNI explanations that are understood by patients is somehow a better explanation, nor does CNI assist the attempts of such charitable clinicians seeking to help their patients to understand. It ought to be possible instead to develop a theory of medical explanation that includes patients and their explanatory needs without relying strongly on a sense of clinicians’ beneficence. How can we engage patients as empowered epistemic agents? This need not require making patients “equal”, because to do so goes against the '25 Not surprisingly, feminist approaches to bioethics have been most active in articulating structural power imbalances in bioethics, and working towards their remedy. 92 common-sense idea of why we seek experts. Instead, it requires a reformulation of the agents of medical explanation. In this chapter, I address the questions: Who are the agents of medical explanations? In particular, what active role should patients play in generating medical explanations? I begin to develop an explanatory schema that sees medical explanation as a collaborative effort, one that empowers patients. This will let us better identify errors in the explanatory process in the cases of Julia, of Uncle Bill and Uncle Boy, and of J .S. (which were discussed in the Introduction). A reconceptualized process of generating medical explanations can be both empowering to patients in a richer sense, and can be a collaborative effort between clinicians and patients. I have argued in previous chapters that the concepts of medical explanation, informed consent, and patient autonomy are intricately tied to our understanding of patients’ decision-making processes. A theory of medical explanation need not be relativistic in order (a) to engage patients as epistemic agents and (b) to identify and respond to patients’ explanatory needs. My goal in reconceptualizing medical explanations is not to eliminate scientific progress or the resulting benefits. Instead, I advance a theory of medical explanation that better describes current explanatory practices that rightly engage patients, and I conceive of alterations that can better improve patients’ experiences. In this way, we can alter ownership, production, and use of medical explanations by putting greater amounts of this process in the hands of patients.126 There are two goals here: one descriptive and one normative. In a descriptive sense, I believe standard theories of medical explanations have failed to recognize the actual participatory roles of patients in developing clinical '26 Laura Purdy makes a similar point in discussing the need to avoid medicalization of certain traits and practices as a goal for a feminist conception of medicine. See “Medicalization, Medical Necessity, and Feminist Medicine”, Bioethics, 15 (2001): 248 — 261 (especially 257). Although Purdy is talking generally of scientific knowledge and theory in medicine (i.e., what is a “healthy” body, or the role of abortion as a medical procedure), her comments apply to the production of explanations as one example of medical knowledge. 93 explanations. In a normative sense, the possibilities for engaging patients in their health care—which I take to be a central tenant of bioethical theory—have not been articulated. Possibilities exist for better explanatory projects that may improve patients’ health, and allow for a greater role for patients in generating medical explanations. Thus, clarifying the possibilities of patients’ roles in the ownership, production, and use of clinical explanations will empower them by making clinical conversations more relevant to their needs and by improving their informed health care decisions. Towards such an end, I draw from Lynn Hankinson Nelson’s “Epistemological Communities”, in which she argues for understanding communities (as opposed to individuals) as primary epistemic agents. I will argue that generating medical explanations is best understood as a group activity, not the activity of individuals. I then will argue that this community view best allows for an improved understanding of patients as epistemically authoritative subjects capable of participating in generating medical explanations. Following that, I will provide ground rules for how such communities should structure their explanations. In this way, I try to recreate how Uncle Bill’s experience could be improved, were his clinician given uptake to Uncle Bill’s explanatory request. Also, this work clarifies the practical components of Nelson’s theoretical framework of communities as primary epistemic agents. Although Nelson wants her theory to reflect real-life practice, she often avoids significant details of how this works. I therefore clarify the role epistemological communities work in medicine, and how this improved description of patients’ roles can be a tool to empowering patients. In this way, I defend Nelson’s theory about the epistemological agents by articulating doctors’ and patients’ roles in generating medical explanations. 94 Gender and Explanations It is important to consider how communities who generate medical explanations are constructed. In the context of discussing medical explanations, it is especially important to note how communities of clinicians, general health caregivers, and patients are constructed. Explanations are tools to improve understanding. But patients and clinicians—like all of us—are socially situated: class, gender, education, race, sexuality, and physical ability all influence our understanding of medicine and how we participate in medical practices. In order to make the point that patients—as individuals and as groups—are socially situated, I want to discuss how gender influences patients’ explanatory needs. This is to illustrate the more general point that medical explanations are generated for the explanatory needs of people with specific social positions. Such social positions have been largely ignored by standard theories of medical explanation. Yet this information is important for understanding the context from which we ask for and respond to why-questions. Here, I would like to briefly consider the influence of gender on medical explanations, and point to areas that need further consideration. Consider Rebecca Dresser’s analysis of the intersection of gender and medical care: A look at the statistics shows that women predominate at all levels of the health care system except those offering power and prestige. “We are 85 percent of all ‘health care’ workers in the hospital, and 75 percent in the system as a whole.” Women enter the health care system as patients more often than men; in addition, women are usually the primary informal caregivers for children, spouses, and aging parents. At the same time, a minority of physicians are women, although the numbers are gradually increasing.1 So, women and men are placed differently within the culture of medicine. As receivers of health care, women are more likely to be patients. As health care providers, women are also situated differently than men. Outside of the clinic, women are more likely then are 127 Rebecca Dresser, “What Bioethics Can Learn from the Women’s Health Movement”, in Feminism and Bioethics: Beyond Reproduction, edited by Susan M. Wolf, 149. 95 men to care for families, adult aging parents, an ill children. Inside the clinic, while women are statistically a majority of care givers, they are in less powerful positions, i.e., less likely to be physicians, and more likely to be nurses. Dresser continues: Most women health caregivers are not physicians, however. Instead, women caregivers tend to be the lowest in the medical hierarchy—“underpaid and undervalued”—and those caring for their children, parents, spouses, and other relatives are not paid at all. Women supply the physical labor to implement “doctors’ orders,” providing burdensome treatment regimes, special diets, and hygienic care. Women most often pay the personal costs of our society’s failure to provide adequate care to elderly and disabled people, frequently giving up employment and other opportunities to care for others. “A familiar pattern is that when a man becomes ill or disabled, his doctor will send him home, commenting that he is lucky to have a wonderful wife to care for him. When the same thin happens to a woman, however, the doctor will recommend a nursing home.”l My point here is not to defend these claims that medical practices are gendered, which Dresser and others have done convincingly in numerous, interdisciplinary ways. I instead take this as a basic background from which to consider the role of and construction of “good” medical explanations: the inquisitors and generators of medical explanations are socially situated beings, rather than abstract, generic, and interchangeable beings. As such, medical knowledge—including medical explanations—ought to articulate how such social situations influence who can know, how we know, and what we can know. Gender, Explanations, and Clinicians Consider the role of gender in clinical explanations such as CNI. While it seems uncontroversial that medical explanations are generated by different categories of clinicians, it is not clear that the explanations generated by these different (clinical) communities will be heard and valued equally. Clinicians are organized hierarchically, and women are more likely to hold positions that are “underpaid and undervalued”, such as nursing. From such undervalued locations, their epistemic credibility will be different than that of their well paid, higher valued male colleagues, e. g., physicians. The point is '28 Rebecca Dresser, “What Bioethics Can Learn from the Women’s Health Movement”. 150. 96 not that medical explanations generated by nurses will be necessarily epistemically inferior. CNI makes no theoretical distinction between explanations generated by doctors and those generated by nurses. CNI’s reliance upon ideal explanations treats the explanation as fixed; the agents of explanation are undifferentiated and interchangeable. Thus social differences between nurses and doctors ought not influence the explanation generated; when it does, it is a mistake to be corrected. Contrary to what the theory of CNI may allow for, there are likely differences between the explanations that different groups of clinicians generate. The point to recognize here is that explanatory communities in medicine are not equal. First, the close and continued care nurses provide may offer them a strong position to understand what is going on with patients they have come to know well; physicians rarely have the time to spend with patients in this way. In this way, nurses may have a privileged stance from which to identify important explanatory information, and have a better sense of what patients’ need in order to understand medical explanations. Patients may also feel they can make different demands from their relationships with nurses than with doctors. Consider a frequently related anecdote: A doctor talks with a patient at the bedside and ask, “Do you have any more questions?” The patient may respond “No.”, sensing that the doctor is busy and has other duties to attend to. But after the doctor has left the room, patients may often ask a nurse to interpret the doctor’s explanation, or to ask the nurse questions that would have been a waste of the doctor’s precious time. Second, and possibly in tension with the first point, nurses’ credibility to generate meaningful knowledge claims is often questioned more than that of doctors. This may have to do with nurses’ shorter medical education, and thus they are perceived as less qualified than doctors to make medical explanations. This does not take into account, though, nurses’ privileged access to knowing about patients through their close interactions. The dismissal of nurses’ explanations may also have as much to do with the social structure of the doctor-nurse relationship. Historically, nurses have been seen as 97 handmaids to the doctor; the doctor is the one who makes medical decisions and the 129 nurse caries out these orders. Thus, the knowledge claims of some groups are often valued higher than others for social reasons, rather than epistemic reasons. Gender Roles, Explanations, and Patients If gender roles are important to recognize in the generation of medical explanations, then they are equally important to recognize in the receiving of medical explanations. As Dresser argues above, women constitute the primary source of health care services in the home, and a significant portion of the entire health care system. Women (as patients themselves) seek out health care services more frequently than do men. As such, women will receive explanations of health and disease differently than do men: more frequently and with greater moral importance to “get it right” on behalf of others. More often than with men, women will be responsible for understanding disease in order to make decisions about treatment choice or how to properly provide care. In the example above, it is easy to imagine that not only does the wife have the responsibility of caring for her “lucky” husband, she has a stronger burden to communicate effectively with the doctor about her husband’s condition. Gender Roles, Explanations, and Topics Warranting Explanations Also, gender and the topic of explanation influence how the explanation is generated. Consider Dresser’s point that clinicians are likely to treat women and men differently for the same condition, whether intentionally or not. In her example, the husband, when ill, is sent home to be cared for by his wife; the wife, when ill, is likely removed from her home and sent to a nursing home to be cared for (likely by other women). Dresser’s point is that clinicians often treat patients differently, even if they are '29 Martin Benjamin and Joy Curtis, Ethics in Nursing, Oxford University Press; 2nd ed edition (August, 1992). See especially Chapter 4 “Reoccurring Ethical Issues in the N urse-Physician Relationship.” 98 not aware of this. While this holds for clinical interactions, do similar differences occur in generating clinical explanations? I argue they do, and this requires correction. The disparities in attention given to men’s health issues over those of women in biomedical research means that in many cases we understand better men’s illnesses and treatment options better than those of women. This problem of unequal research may cause clinicians to provide different explanations: clinical explanations of men’s health and disease are more readily generated. Consider that heart disease remains under- diagnosed and under-treated in women, despite the fact that heart disease is the leading cause of death of American women. Women are also more likely to be diagnosed with “ambiguous” conditions that defy easy diagnosis and explanation.130 Explaining women’s health and disease may often be briefer because the proper research is unavailable as a foundation for the explanation. '31 In other cases, clinicians may be unaware of the problem that requires an explanation, or unclear which of a number of possible explanations is appropriate. Consider the case of J .S., and her series of vague and unenlightening diagnoses, which was discussed in Chapter 1. Her doctor might take seriously that how women present disease is connected to their social situation, e. g., exposure to abuse or stress. While Thagard’s theory emphasizes explanations as causal connections, such a model is not suited for a wide range of explanations. J .S.’s medical condition is an example of what has come to be called “medically unexplained disorders.” Kirsti Malterud, a physician, has described such disorders in her own writing, and in collaborative writings with Lucy Candib, another clinician, and Lorraine Code, a feminist philosopher. '30 Anne Werner and Kristi Malterund, “It is Hard Work Behaving as a Credible Patient: Encounters between Women with Chronic Pain and Their Doctors” Social Science and Medicine, 57 (2003): 1409- 1419. '3‘ Notice this argument is consistent with Thagard’s original claims that “clinical explanations” are derived from “research explanations.” If there are problems with how biomedical research has been done, then this will adversely affect clinical explanations. 99 Malterud describes medically unexplained disorders as “chronic and disabling conditions, presenting with extensive subjective symptoms, although objective findings or causal explanations are lacking.”132 A few common examples might be fibromyalgia, chronic fatigue syndrome, irritable bowel syndrome, or temporomandibular joint disorder (TMJ). Such conditions occur mostly in women, although occasionally in men. Also, patients with unexplained medical disorders are more likely to report histories of sexual or physical violence.133 Much of the literature on unexplained medical disorders involves the frustrations that arise when trying to incorporate social determinants of health as explanatory. Thagard makes no allowances for this sort of explanation. Again, the problem is that with such disorders, there is no clear causal connection that can be clearly referred to in CNI. And without clear causal connections required for CNI, no explanation can be generated. Clinicians and patients involved with “unexplained” medical disorders face a number of pressures to provide a diagnosis, even if it does not “fit well”. For example, something needs to be recorded on medical charts; insurance agencies may not reimburse unless a diagnostic label is provided. Yet this uncertainty of diagnosis is often not revealed to the woman herself (causing her to feel unheard or ‘crazy’ for thinking such things). She may be given unhelpful treatments and conflicting advice from multiple clinicians, and eventually feel the frustrations echoed in J .S.’s case.134 J .S., as a patient, is doubly burdened by both illness and also by a range of social pressures due to her gender. Consider what pressures she faces at home: likely she is caring for her ailing mother—in-law, without the help of her husband, and she does this while maintaining her job at the hospital, and while feeling ill herself much of the time. '32 Kirsti Malterud, “Symptoms as a Source of Medical Knowledge: Understanding Medically Unexplained Disorders in Women”, 604. '33 Malterud, “Symptoms as a Source of Medical Knowledge”. See also Alexandra Ilnyckyj and Charles N Bernstein, “Sexual Abuse in Irritable Bowel Syndrome: To Ask or Not to Ask—That is the Question,” Canadian Journal of Gastroenterology, 16 (2002): 801-805. '34 See also Kristi Malterund, Lucy Candib, and Lorraine Code, “Responsible and Responsive Knowing in Medical Diagnosis: The Medical Gaze Revisited,” 8-19. 100 The fact that she is a woman who is facing the forces of social oppression related to her gender ought to be in some way explanatory of her health problems.135 The closest Thagard comes is allowing for vague terms such as “stress”. But it seems incorrect to describe the complex problems of the patient J .S. as only the result of “stress”. Consider just one piece of her situation, the fact that J .S. has taken on the significant role of care giver for her mother-in-law. Such responsibilities to care for the ill often fall to women. To describe her burden as “stress” misses the larger story, the complexities of her situation, and the way social forces are contributing to her illness.136 By this omission, Thagard’s schema re-inscribes a common problem: his explanatory schema has as its goal simplifying explanations. This allows clinicians to gloss over problematic information for the sake of simplicity, rather than develop clinical explanations that are complicated and perhaps even, at times, contradictory. In this case, the clinician likely will try to describe J .S.’s problem as psychological or physical. Standard medical explanations often reduce the complexities, rather than generate them. Instead, it may be that a good explanation of unexplained medical disorders should seek to keep explanatory gaps open, and to keep us asking additional questions about underlying causes.'37 In this way, a good clinical explanation is more likely to account for the complex interactions of physical, psychological, and the social determinants that resulting in J .S.’s problems, rather than focus on any one over the '35 I use the term “oppression” in the sense developed by Marilyn Frye: oppression is a social force affecting groups, not individuals. So J .S. suffers oppression as a woman, not as an individual. Women generally are socially situated such that they are pressured to be the family nurturer/caretaker. See Marilyn Frye, “Oppression,” The Politics of Reality (Freedom, CA: The Crossing Press, 1983) 2. ‘36 Typically, I want to keep “explanation of disease” and “treatment of disease” separated as different medical activities. Having said that, if J.S. is suffering from stress, her doctor may prescribe medicines like a relaxant. or suggest she take a yoga class. Yet even if these reduce her stress, it likely misses what is the real (or deeper, or distal) cause of her health problems. For instance, it may be previous sexual assault, and labeling the effects as “stress” never lets us see this cause in the proper light. 137 This is not necessarily in conflict with Kim’s view of what explanations do: to relieve epistemic puzzlement (see Chapter 1). If a clear causal explanation (like what Thagard wants) is unavailable, a discussion of what may be the cause(s) is often better for the patient than denying that certain features are the cause. It may better to for clinicians to admit, “Well, I don’t know, it may be X, it may be Y, or some other 2”, rather than say “Well, it’s not physical, so it must be psychological.” There may be degrees of relieving epistemic puzzlement that allow for multiple possible options to be considered at once. 101 others. A good explanation may also describe the areas where questions about causation remain, rather than try to force these conditions to fit under other diagnostic labels. Explanations are requested, generated, evaluated, and received by persons who are embodied, raced, and gendered, rather than by disembodied entities utilizing a view- from-nowhere. In the clinic, specific persons generate medical explanations about the health or disease of specific persons; again, such explanations are contextual in important ways. The point here is that gender is tied in important ways to what is considered a “good” explanation. First, the gender of the explainer matters in how we generate medical explanations. Second, the gender of the inquisitor and receiver of medical explanations matters. Given women’s role in providing health care, they are more likely to be questioners and receivers of medical explanations. Third, gender influences which problems are identified as requiring explanations, as well as the degree of concern about providing a “good” explanation. Therefore, a discussion of “good” explanations ought to include whether a theory of medical explanation adequately meets the needs of women as health caregivers (broadly conceived). Yet the social situatedness of explanation inquisitors and responders has been for the most part ignored by discussions of scientific explanations generally, and medical explanations in particular. Here, I claim only to have addressed the beginnings of this conversation about the gendered context of medical explanations, and more work remains to be done on this. In the rest of this chapter I argue for one strategy: analyzing the social situatedness of explanatory agents. I draw from the work of Lynn Hankinson Nelson to argue that the primary agents of explanations are best understood as communities, and derivatively as individuals. Again, I take it that most theories of medical explanation and scientific explanation have overlooked the strong situatedness of explanatory participants. In the section above, I have begun a description of how gender plays an important role in how explanations are requested, generated, and received. In the rest of the chapter, I will look 102 more generally at how the agents of explanations are socially situated. In the next chapter, I continue this by noting how explanatory interests may be accounted for by van Fraassen’s pragmatics of explanation. This leads us to the question discussed in the following section: who are the agents of explanation? Reconsidering the Agents of Explanation Medical explanations, I have argued in the previous chapters, are tools that are generated for certain purposes, rather than, say, discovered aspects of the world. But who are the agents who generate them? At first glance, clinicians generate them. This is the position that has been taken up largely by Thagard, although he can be read to understand clinicians and biomedical researchers in groups as well as individuals as explanatory agents. The point is that he remains ambivalent in establishing who the agents of medical explanation are and what their respective roles are.138 Both the clinician and biomedical scientists are described as explicating an already established (ideal) medical explanation. For now, I shall set aside the proper role of clinicians in regards to generating medical explanations. We are still left to ask what role, if any, do patients play in medical explanations? Thagard gives no clear indication to how patients develop (or collaboratively help to develop) medical explanations. But does this mean patients should be understood as having no such role? I argue instead that the role of patients has been overlooked. Consider the case above. Uncle Bill asks his doctor for an explanation of his stroke. Uncle Bill’s role, at a minimum, is to ask for the explanation. In other words, patients’ roles in explanation have been typically limited to that of requesting the '38 At times, Thagard’s examples indicate that clinicians generate the explanations (see the example of Julia’s ulcer, How Scientists Explain Disease, 15, which I also discuss in Chapters 1 and 3). In later chapters of How Scientists Explain Disease, Thagard is primarily interested in the social network and practices that allow biomedical researchers to generate medical explanations (for example, Chapter 8 “Discovering Causes”, Chapter 11 “Collaborative Knowledge”, Chapter 12 “Medical Consensus”). But there is a fluidity in these discussions such that it is often unclear if individual clinicians can (or cannot) provide explanations differently than might the collaborative group of all biomedical researchers. This makes the clinician, as an explanatory agent, often synonymous with current scientific consensus rather than an independent generator of medical explanations. In this way, the explanatory agent as “scientific community” can easily be grafted onto Thagard’s CNI. 103 explanation. Clinicians then have the “active” role of generating the explanatory response and providing it to patients.139 Thagard, though, does not address the potential active role of patients. I have argued that we must move beyond this limited scope so as to include patients’ role in this process. Yet this component of medical explanations has been undertheorized. So, this concept of the agents of explanation needs unpacking. For instance, why is the standard strategy of medical explanations—clinicians generate them and bestow them on patients like Uncle Boy—problematic? Although it is often a means of disempowering patients, this is not its only failing. These standard strategies of medical explanations also misunderstand how medical explanations are generated. Medical explanations are best understood as projects of groups rather than of individuals; these groups are often comprised of patients and clinicians, of scientific lay- persons and scientific experts, all of whom gather together in a variety of numbers and groupings. Communities as Knowers In this next section, I will utilize a conception of epistemic agents as communities or groups, rather than individuals. The rejection of the solipsistic knower is common in much of feminist philosophy of science and epistemology. ”0 Lynn Hankinson Nelson’s provides one of the clearest such explications.141 Nelson’s work, which draws heavily on Quine, emphasizes how communities (not individuals, who are abstractable members of '39 Again, the clinician here is the stand-in for the greater scientific community in generating medical explanations. Any clinician could do the same work. 14° See, for example, Sandra Harding’s development of standpoint epistemology in Whose Science? Whose Knowledge ?: Thinking From Women ’s Lives (Ithaca: Cornell University Press, 1991) and “Rethinking Standpoint Epistemology: What Is Strong Objectivity?’ in Linda Alcoff and Elizabeth Potter (eds), Feminist Epistemologies (New York: Routledge, 1993). See as well Helen Longino’s argument for the value of pluralism in feminist epistemology in Science as Social Knowledge (Princeton: Princeton University Press, 1990). '4' Lynn Hankinson Nelson, “Epistemological Communities” in Alcoff and Potter, Feminist Epistemologies, 121-159. 104 such communities) engage in projects of gathering evidence and justification.142 Nelson writes, it is in and through a variety of such activities that knowledge is generated. The change I am proposing involves what we should construe as the agents of these activities. My arguments suggest that the collaborators, the consensus achievers, and, in more general terms, the agents who generate knowledge are communities and subcommunities, not individuals.143 Communities as knowers, according to Nelson, are thus epistemically prior to individuals as knowers; individuals know, but only as derivative of what communities know. Generally, feminist epistemologists have argued that the social nature of knowledge (and of knowing) needs further clarification and a more central role in theories.144 Nelson acknowledges “three assumptions” motivating her elaboration of this view of communities as epistemic agents: “That the category ‘agents of epistemology’ is dynamic; that our views of such agents are interdependent with our understandings of evidence; and that epistemology is radically interdependent with other knowledge and projects.”145 In a broader sense, though, Nelson (as well as other feminist epistemologists) find the classical epistemic subject of Descartes incompatible with other claims and theories of feminist epistemology.146 As Heidi E. Grasswick writes, ”2 Lynn Hankinson Nelson, Who Knows: From Quine to a Feminist Empiricism (Philadelphia: Temple University Press, 1990). ”3 Nelson, “Epistemological Communities”, 124. :44 It also should be asked: what is particularly feminist about Nelson’s work (and that of others)? I like the following list of four criteria that all feminist projects share. Although it is not a neat definition, it seems likely that nearly all feminists share these basic four assumptions as beginning points for their projects. 1- Women are oppressed. 2-This is not a natural, inevitable, or unavoidable phenomenon. 3-Oppression in all forms is wrong. 4-It is possible to alter our social order and create societies that are not oppressive. In this sense feminists are committed to improving the lives of women. Of course, the eradication of oppression means that not only women will benefit from feminist change. I believe these four points were originally articulated by Alison Jaggar. I thank Jennifer Benson and Allison Wolf for calling this to my attention. ”5 Nelson, “Epistemological Communities” 125. '46 Note, it may be possible to connect medical explanations with the work on feminist standpoint theory. But I intentionally have not done so here. In seeking the goal of a collaborative, non-oppressive explanatory project, standpoint theory is an interesting beginning point in that it acknowledges explicitly from the outset the differences between patients and clinicians in situatedness and power relations. Yet merging this gap (of power and of understanding) between disempowered patients and empowered 105 it is difficult, if not impossible, to incorporate many of the claims of feminist epistemologists concerning the situatedness of knowing, the influence of power relations between knowers, and the various characteristics of our communities that shape the construction of knowledge, while remaining committed to the idea of the epistemic subject as an atomistic, self-sufficient individual.147 Thus, the standard epistemic agent often is inadequate to describe who knows and how we know in some projects. Thus, it is descriptively inadequate, as well as limiting our abilities to make normative improvements. The atomistic epistemic agent is not adequate for an accurate description of knowledge projects involving the interaction between multiple communities. Much of feminist knowledge has not been generated independently, but as part of collaborations. For instance, many feminist philosophers of science have developed critiques of scientific theory and practice—identifying power and gender relations—precisely because of “these critics’ participation in both feminist communities and (nonfeminist) science communities, indicating that knowledge developed precisely because these communities were not isolated and self-sufficient.”148 Thus, advancements in feminist theory regarding critiques of science are largely due to the collaboration of feminist philosophers and scientists. At the broadest level, this includes the general community of feminist critics of science. At the narrow level, this can occur in the collaboration between individuals, say, between a biologist and feminist philosopher who come together to try to understand how gender and power are shaping the research practices of the biologist. '49 clinicians, I believe, makes it difficult beginning theory from which to develop a collaborative project between clinicians and patients. “7 Heidi E. Grasswick, “Individuals-in-Communities: The Search for a Feminist Model of Epistemic Subjects” Hypatia l9. 3 (2004) 85-120. ”8 Grasswick, “Individuals-in-Communities” 93. '49 This example comes from a presentation at the Association for Feminist Epistemology, Metaphysics. Methodology and Science Studies (FEMMSS) conference, 2004. 106 While Descartes’ epistemic subject was “basically passive, a recipient or collector of knowledge,”150 Nelson’s active epistemic communities “shape, as well as undergo and absorb, experience; they construct meaning and knowledge, and even. . .negotiate and decide these.”15l These communities are historically situated, a situatedness that influences how they consider standards for evidence. Modern scientific knowledge is generated by the scientific community’s commitment to practices like the scientific method and publication and peer review of data. But science could have developed differently and utilized different practices to generate knowledge. As Nelson writes, “standards of evidence are historically relative and dynamic, emerging concomitantly with the processes through which knowledge is generated, rather than having been laid down prior to these processes.”152 The idea of changing standards of evidence is not unique to Nelson (or to Quine),153 but Nelson’s emphasis here is that the community of science (rather than individuals) collaborates and agrees upon these changes. Nelson’s main point is that communities, not individuals, should be considered as primary epistemic agents.”4 What role, though, does this leave for individuals as the constitutive members of epistemic groups? Nelson writes, “although I do not think individuals are the primary epistemological agents. . .I do not deny that individuals know. '50 Or, as Grasswick summarizes the “atomistic agent”, such knowers are “individual, self-sufficient, and generic”. It is because of “his” endowed faculties (reason, sense perception, which are universally shared) that he can come to know the (passive) objects of his attention (“Individuals-in-Communities” 89). ‘5' Nelson, “Epistemological Communities” 121. '52 Nelson, “Epistemological Communities” 122. ”3 Although Nelson does not cite him in this article, Thomas Kuhn (The Structure of Scientific Revolutions, (Chicago: University of Chicago Press, 1962)) argues that there is no formula to dictate which theory we should utilize from a set of competing theories. Advancing this position, Larry Laudan (Progress and Its Problems: Towards a Theory of Scientific Growth (Berkeley; University of California Press, 1977)), argues that standards of evidence can change over time, and this can be a rational process. '5‘ For critiques of Nelson’s concept of epistemic communities, see Mark Owen Webb “Feminist Epistemology and the Extent of the Social,” Hypatia 10.3 (1995): 85-98. He argues by analogy that the concept of the community as primary knower fails. See also Grasswick, “Individuals-in-Communities.” Rather than arguing that epistemic communities as a concept fail, and thus should be discarded, Grasswick seeks a complementary solution. She argues for understanding individuals as primary epistemic subjects, but individuals who are different from solipsistic knowers. These individuals are members of knowing, learning communities, and these individuals are aware of the dynamic reciprocity of their epistemic activities. Although I do not take up Grasswick’s critiques and suggested improvements in any detail, I find her consideration of this topic thoughtful and thought-provoking. 107 My claim is that the knowing we do as individuals is derivative, that your knowing or mine depends on our knowing, for some ‘we.”"55 Thus for Nelson, epistemic communities as knowers are ontologically prior to individuals as knowers. Even when we know as individuals, we “do not ‘dissolve’ into ‘collections’ of knowing individuals.”'56 We as scientists (or nurses, or teachers, or feminists) remain members of epistemic groups and retain our connection to prior epistemic groups, and our knowledge is not something that no other individual could have achieved. How do various communities and subcommunities collaborate in order to generate and share knowledge? Nelson writes: epistemological communities are multiple, historically contingent, and dynamic: they have fuzzy, often overlapping boundaries; they evolve, dissolve, and recombine; and they have a variety of “purposes” and projects which may include (as in the case of science communities) but frequently do not include (as a priority) the production of knowledge. ‘57 She continues: communities are dynamic and unstable. They evolve, disband, realign, and cohere as interests and undertakings evolve and are abandoned, as new experiences, standards, and knowledges become possible. . ..There are subcommunities that have developed categories, methods, projects, knowledge, and standards in addition to those they share with larger communities (e. g., the physics community is a subcommunity of a larger community with which it shares knowledge and standards: a community on which it is, in several senses, dependent). There are also subcommunities that have generated knowledge and standards that challenge aspects of a larger body of shared knowledge and standards. Some examples of these are the various subcommunities of feminist philosophers, communities that share some (but not all) of the knowledge and standards of the community of philosophers.‘58 ”5 Nelson, “Epistemological Communities” 124. '56 Nelson, “Epistemological Communities” 150. '57 Nelson, “Epistemological Communities” 125. '58 Nelson, “Epistemological Communities” 148- 149. 108 Some communities traverse “more than one discipline or science,” and a specific community can influence another. 159 Thus, a community generates knowledge based on the agreement about epistemic practices, e.g., how to gather and judge evidence or what theories and background knowledge to accept (or disregard). Such practices need not be universal, although different communities may adopt similar practices, or draw from the practices of shared prior communities.160 Consider the phrases, “I know that the nucleus of the atom is divisible,” or “a sedentary lifestyle is unhealthy.” For instance, I, personally, have no direct experience with the nucleus of atoms. In making claims about the atom, I am strongly relying upon the knowledge of others, persons who I consider to be experts (or more knowledgeable than am I) about atomic structures. And, it should be noted, even these experts do not have direct sensory experiences of atomic structure. So, when I discuss atoms, protons, and neutrons I am largely pulling from the accepted knowledge of a community that has agreed to certain scientific practices, and I understand (whether or not I am directly considering it at this time) that I am connected to a certain community of knowers somehow. I have a general understanding of and historical engagement with scientific practices, which has lead me generally to trust such findings. My knowledge claims are not due to my direct observations, which could be another alternative. Instead, I am a member of a group that has relied upon scientific practices. From those scientific activities, I have come to accept that the nucleus of that atom is made up of smaller pieces. So in summary, it makes sense to say that the group of knowers (to which I may belong) exists prior to my knowing. '59 Lynn Hankinson Nelson, “Empiricism Without Dogmas”, in Lynn Hankinson Nelson and Jack Nelson (eds) Feminism, Science, and the Philosophy of Science (Boston: Kluwer Academic Publishers, 1997) 103. '60 In Kuhn’s sense, each new paradigm will allow for a set of standards for evaluation of evidence. What is interesting in Nelson’s strategy is there may be multiple standards at a given time for evaluating the same evidence, whereas Kuhn really only allows for the current ruling paradigm’s standards. 109 This leads to the question: what sort of group can constitute an epistemological community? The concept of a ‘community’ as an epistemic agent remains vague, and its borders remain difficult to pin down precisely. Much of this is intentional on Nelson’s part. Nelson argues there are not strict boundaries or requirements for epistemic communities. They are dynamic, rather than monolithic or stable. [C]ommunities and their parameters will be a function of the nature of our projects and purposes. . .; of the definitions communities give to themselves and the projects they undertake; and of the importance such communities. . .attribute to the standards and knowledge they share with larger groups and those they do not—decisions which will also be relative to specific purposes and interests.‘6' What is of importance in understanding the makeup of an epistemic community is how they set and share standards and practices for generating and evaluating knowledge. For Nelson, this is particularly important for showing how feminist knowledge is possible. For instance, feminist phiIOSOphy critiques other academic disciplines, including non- feminist philosophy and science. In this way, feminist philosophers reach beyond disciplinary boundaries to engage with other types of knowing. Under this model of collaborative epistemic agents, epistemic projects are generated by, and evaluated by, the communities where the knowledge- generating occurs. Although Nelson does not argue for epistemic communities this way, one might think about a ‘community’ in terms of a Wittgensteinian family resemblance. Such communities do not share any single essential features, but there are a number of features they may share. For example, size will be an issue. Much like social senses of communities, epistemic communities can vary in size, from a pair to a vast collective, yet an individual by herself cannot constitute a community. Shared interests and goals will determine membership. The choice of knowledge projects undertaken, and what guidelines will be followed to evaluate information will be negotiated by the group. "" Nelson, “Epistemological Communities” 149. 110 When disagreements arise, the community can resolve the differences, or the community can disband completely, or redistribute into new communities. The members of the community may have different roles or jobs; occasionally these roles may come into conflict. There will be boundaries to membership: not all persons will be part of any one community. There will be some shared temporal history to the community and its 9 activities. Finally, epistemological communities generate knowledge claims that are publicly shared, and thus open to critique and comments of others.162 While not all epistemological communities will share all of these features, each community will generally have many of them. Communities Involved in Medical Explanations I believe that Nelson’s description of epistemic communities allows us to understand better the complexities of how medical explanations are developed: how research explanations (shared between researchers) are both similar yet different from clinical explanations (those shared between doctor and patient). Research explanations are generated by an epistemic community that generally shares an agreed upon set of practices. Yet clinical explanations involve agents who come from diverse backgrounds. Patients and clinicians must therefore develop shared grounds for generating and evaluating medical explanations. Additionally, the theoretical basis of the epistemic community allows for an improved sense of how clinical explanations ought to be generated in ways that can empower patients as epistemic agents. There are three points I want to make about medical explanations and epistemic communities. First, medical knowledge generally can be understood as primarily derived by a medical community, and secondarily by individual clinicians. In other words, that the medical community is a prior epistemic agent, and the individual clinician or '62 For instance, science has the process of peer review. Other communities may share findings through less formal methods, such as narratives. 111 researcher is a derivative epistemic agent. Second, patients comprise a parallel epistemic community, one that is engaged in valuable medical epistemic projects. Third, communities of clinicians and patients collaborate to generate clinical explanations. Again, it is first as a community that doctors and patients collaborate to generate medical knowledge, including medical explanations. Only derivatively do they, as individuals or as small teams, generate clinical explanations of an individual case. Clinicians as an epistemic community Much of Thagard’s work in How Scientists Explain Disease deals with the many procedures, methods, language types, categories, patterns of reasoning, and practices, as well as specific sub-groups that scientists develop as part of biomedical research, such as theory justification, peer-review, and publication of data. In this way, scientists remain a community of their own, which is part of a larger community of knowers. Scientists create sub-groupings, according to their disciplines and according to projects (e.g., peer- review of data and research methods, articles, and grand proposals). I do not think Nelson’s arguments are incompatible in any strong sense with large portions of Thagard’s conception of scientists (e. g., as a community or as developing medical explanations), despite the fact that Thagard does not draw upon this literature or share in Nelson’s beginning assumptions or her political desiderata. Thagard may disagree about the primacy of communities over individual knowers, but he does discuss much of the social mechanics that underlie biomedical research, which in turn give rise to medical explanations. Clinicians and biomedical researchers—as a knowing community—fit many of the aspects of Nelson’s epistemic community. They do not comprise a necessarily fixed or stable group. Any one clinician may be a member of multiple groups, both with and without medical interests: the group of practicing clinicians, of biomedical researchers, of political activists, of their own families. Even in focusing on their medical projects, 112 knowledge is generated as part of scientific research. A different type of knowledge is generated at the bedside: the skill of interacting well with patients. In both situations, the choice of topic and research program will focus clinicians’ interests and specializations. Explanations of disease play a role in both clinical and research aspects of doctors’ activities, which is Thagard’s main point of interest. Clinicians produce valuable types of knowledge both as researchers and as clinicians. Patients as an epistemic community Thagard’s explanatory schema does not recognize patients’ roles as epistemic agents. This omission is telling in that it prevents, in large part, a theoretical toehold for certain critiques; the power dynamics between patient and clinicians are erased. I believe Nelson’s concept of epistemic communities is helpful here to better articulate (a) how “patients” as a community contribute to generating medical explanations, and (b) how failure to recognize patients as epistemic agents often leads to failed clinical explanations. Thagard and others writing on medical explanation have overlooked this point, and it remains insufficiently developed. It must be noted that the community of ‘patients’ is not monolithic. There will be great variation in this community when examined in greater detail. Just as the community “scientists” can be divided (or reunited) according to their interests in a given project (e.g., the differences between biologists, chemists, and physicists, or feminist-scientists versus non-feminist scientists), so to the community of patients is divisible and flexible. For much of my discussion, I may generally talk of “patients” as having roughly similar interests and as in contrast to the interests of clinicians. Yet patients’ interests will vary dramatically when viewed upon the axes of gender, sexuality, race, socioeconomic status, age, historical and geographical setting, physical ability, or severity of illness, or type of illness. 113 Patients as community of knowers: This leads to the more controversial of my arguments, that patients are an epistemic community: knowers in their own right, who generate knowledge that can provide valuable information to clinical practice. Patients are communities of knowers, too. The community of patients is comprised of individuals who are, likewise, members of families, political groups, and the larger scientific-lay public. Patients also comprise a community of persons connected by their experience and understanding of illness, pain, loss of bodily function, and engagement with medical communities and institutions. The community of “patients” also includes clinicians as they become ill, especially in areas outside their expertise.'63 As a group, patients interact in numerous ways with clinicians to generate knowledge, and part of this process is the generation of medical explanations. For example, a community may form around people who share a given disease. In the 19803 and 19905, AIDS activists generated a community of patients who became highly engaged with their medical care; political groups like ACT-UP developed as a means to lobby on behalf of persons with HIV/AIDS. Patients with HIV/AIDS often had to manage their own care in important ways, since little was known about the disease. Over time, as more was understood about the disease, patients organized to seek expedited FDA approval for AIDS drugs. Support groups emerged where patients shared information about the disease, about which treatments worked, and about which clinicians took their care seriously. Elderly persons may serve as another example of a medical community. Retired and elderly persons make up a significant political force; consider the important role of AARP in rallying for political consideration of healthcare benefits for retired persons. '63 Some might think that clinicians, even when sick, act more like clinicians than scientific-lay patients. For an interesting discussion of why this often is not the case, one clinician describes her serious difficulties navigating her experience as a cancer patient. See: Anna Donald, “Balancing the Benefits and Harms of Care” British Medical Journal 329 (2004): 59. I thank Libby Bogdan-Lovis for calling this to my attention. 114 More recently, elderly persons have lobbied for improved drug coverage by insurance agencies. When such coverage was not forthcoming, bus trips to Canada and Mexico and mail order systems were developed as a means to find cheaper sources for prescription drugs. Child birthing classes are another example of patients acting as a community. There is a long tradition of women teaching other women about their bodies, about the birthing process, and about what to expect while giving birth in a hospital. Birthing coaches often work with women to explore their goals for the birth, and to develop a plan for the birthing experience they are seeking (for example, giving birth at home versus in a hospital). As another example, consider the case of Uncle Bill from Chapter 1. At the first level of patient involvement, Uncle Bill asked the why-question to his physician, “Why did I have a stroke?” The resulting response was, to Uncle Bill, inadequate. And, it is questionable whether the clinician took the why—question (or the generation of an explanatory response) seriously. In what way is Uncle Bill and Uncle Boy part of an epistemological community of patients? Uncle Boy clearly outlines how power, race, and wealth work in his town. Although he may have observed instances of this first hand, Uncle Boy has not come to understand how racism works all by himself. Instead, he has been raised and educated in a particular community, one where black folks come to understand how white folks live their lives.“54 Beyond an understanding that racism exists, Uncle Bill and Uncle Boy have knowledge of how this racism adversely affects black folks’ health. This general '6‘ Standpoint theorists, especially Patricia Hill Collins, have made this point. Black folks, to ensure their own survival, have had to come to understand the lives of white folk; white folks have not had to take on the project of understanding themselves for their own survival. See Collins’s “The Social Construction of Black Feminist Thought,” Signs 14. 4 (1989): 745-773. A stronger point is made by Charles Mills in The Racial Contract (Ithica: Cornell University Press, 1999). He points to an “epistemology of ignorance”, where he argues that White folks have created the world they occupy, a world in which they cannot come to understand their role in it. 115 relationship between racism and poor health seems well understood by Uncle Boy, yet his clinician may not have indirect or direct knowledge of it (especially if, say, he does not understand racialized nuances of his life, and clinical trials have not been performed that correlate race and health disparities). Uncle Bill, though, has a greater understanding of what it is like to be a poor, black man in this town. If we can assume he has had experiences similar to those of his brother, Uncle Boy, then he has an understanding of how power differences work in issues of race and class. It is unclear whether the doctor in this case understands this, as well. Although it is possible he does see these power differences, he likely is ignorant of them.‘65 As the primary doctor for this factory, he may be an important cog in this machine that works to disempower the employees of the factory, but be unaware of his role as such. Uncle Bill and Uncle Boy, though, have a clear understanding of the power positions of the doctor, of the factory owners, and of themselves.“56 '65 This is not to say that all clinicians are unaware of social forces like poverty, racism, sexism, or homophobia. This clinician, I think it is fair to say, is thusly ignorant. While raising these topics in medical education is one solution, it is often received by medical students as “social” stuff, and not real medicine. I think this attitude, while mistaken, is also common. '66 Note here, much work could be done with feminist standpoint theory. It is easier for Uncle Bill and Uncle Boy to understand their position, because they are at a disempowered position. Consider the importance of epistemic positions in standpoint theory. As Harding and others (e.g., Patricia Hill Collins, “The Social Construction of Black Feminist Thought”) have argued, persons lower on the power-ladder have an epistemically privileged view. They are better positioned to understand the lives of persons higher on the ladder, those persons with a greater social privilege. Day-to-day survival for persons lower on the power ladder requires that they understand those persons higher on the ladder; yet there is not equivalent need for survival that requires privileged persons to understand themselves. Consider this brief example: a maid at a luxury hotel must know the lives of the guests she serves. Yet if she does her job well, patrons of the hotel will not know that she exists, let alone will they understand her political, social, and economic situation. The imbalance of this relationship requires that the socially privileged be understood but the less privileged, and that the privileged have barriers of various sorts that keep them from understanding the lives of the less privileged. This is to make the point in this case that it will be beneficial yet difficult, although not impossible, for the clinician to understand his own role in oppressing his patients. He can actually think he is benefiting them, when in fact he is harming patients in ways he is unaware of. Yet because Uncle Bill and Uncle Boy have different social positions, they have a clearer “view” of the doctor, and thus are better aware of him than he is of himself. So, it is important to note the adequacy and deficiency of some epistemic communities. Some communities fail to dredge up all of the relevant information, and this failure can be for a number of reasons. In the case of this clinician and Western medicine generally, their failure to learn from underprivileged standpoints make them deficient. 116 While the clinician may choose not to give Uncle Bill uptake on his knowledge, this conception of the patient begins by assuming that patients are knowers of their own sort. Thus, clinicians (as one type of knower) can engage with patients (a different type of knower). The benefit is that this assumes patients, from the outset, are likely epistemically valuable agents in their own right, and thus are capable to be participants in medical explanations, rather than requiring the patient to prove that she is worthy of being taken seriously. The clinician-patient team as epistemic community Understanding communities as epistemic agents allows a wider, more fruitful sense of doctor-patient encounters, the nexus where clinical explanations are generated. I will argue that the clinician-patient community is the primary agent of clinical explanations. Such engagements form new yet flexible groups—a new, temporary community that is informed by a combination of their previous groups’ beliefs (from lay and scientific background). I will consider at least four ways patients can be involved in medical explanations (from weak to strong engagement): the patient as inquisitor; the patient as providing contextual clues; the patient as providing personal narratives of significance; and clinicians-patients as a community of knowers. This is not an exhaustive list, but the first three are common themes that arise in the literature on doctor-patient communication. I will argue the fourth option—the interactions of clinicians and patients as an epistemological community—best meets the needs for developing clinical explanations. Patient as inquisitor: At the most basic level, the patient’s role can be conceived the asker of why- questions. In such a case, the patient is active in the sense that she initiates the request for an explanatory response. The patient remains only weakly active as an epistemic agent at this level. Although the patient is an inquisitor, she plays no role in generating or 117 evaluating the explanatory response.“57 Largely, this is Julia’s level of participation with her doctor regarding her ulcer. Although Julia asks a question about her condition, the clinician generates the explanation. It is at this level, too, that I argue Thagard believes clinicians generate Cle as a response to clinical explanations. While this may be technically an interactive explanatory process between clinician and patient, it is interactive in only a weak sense. The clinician is the primary agent of explanation. A slightly “stronger” conception involves reiterating the questioning process. As a patient asks one question, the doctor likely provides a response. In light of this response, though, the patient or the doctor can ask a new or refined question. This is to make the point clearer that the question-and-answer aspects of doctor-patient conversations are often complex dialogues or exchanges, rather than a simple, well-formed question with a well-considered response. Patient as providing contextual clues: Patients may be the source of numerous contextual variables. For instance, the clinician can metaphorically “read” the patient to gain a richer sense of situation: What is the patient really asking about? What is the best way to respond? In this way, the patient is a sort of index of contextual features, many of which the clinician will (intentionally or unintentionally) draw upon as she shapes the conversation and the resulting treatment plan.168 Although the patient is the source of such contextual knowledge, clinicians take much of it up without explicit effort on the part of the patient. Clinicians, though, can know the importance of this information, and be trained in how to best gather it.“59 At this second level, Uncle Bill has knowledge of where he has been, what chemicals he has '67 These three levels are not a complete taxonomy of possible means of patient participation. I intend these examples to show a few importantly different means of engaging in explanation. Also, these are possible— not mandatory—roles for patients. '68 In the next chapter, I develop in greater length how the clinician can identify and utilize these contextual features in generating and evaluating medical explanations. ‘69 For one example of this training that explicitly addresses how clinicians can try to determine to what extent a patient understands his or her situation, see “Breaking Bad News: A Six-Step Protocol” in Robert Buckman, How to Break Bad News: A Guide for Health Care Professionals. 118 been exposed to at the workplace, how often has this occurred, and what preventative and safety measures are in place at the factory (if any). These, assumedly, are facts that would be clinically useful to his physician in explaining the stroke. Patient as providing personal narratives: At this level, patients may serve as inquisitors, as described in the previous section. Going beyond this, though, patients also provide information of their own experiences to the clinician, that is, personal narratives of significance about their individual experiences. Much of this is information garnered during the taking of medical histories. This may be information that is not directly implied by the question. The story is often a beginning to the explanatory process. The patient provides information that she believes is important for the doctor to understand, and other information that may or may not be relevant. Here, the patient is active in that she provides the information that the clinician will likely analyze and from which generate an explanation. For example, one woman with an “unexplained medical disorder” (chronic neck pain) explains to her physician, “I carry my stress in my neck.” Another woman explains to her doctor that her heart condition (early stage congestive heart failure) is directly linked to the physical abuse she suffered from her husband. In providing these narrative explanations, it may be that in some ways, the patient does a bit more work: by her determining which narratives to provide, she shapes what information the physician will consider. At this level, though, the clinician remains the primary epistemic agent, the person who decides what information is valuable (and which information should be disregarded). Because there is little data available, clinicians may not find the connection between abuse and heart disease reliable. Although a patient may know she carries stress in her neck, her clinician may explain her condition by referencing disk degeneration, that is, by generating a physiological explanation rather than one that instead, or in addition, cites psychological conditions. The clinician may also infer from the narrative pragmatic and contextual features that may be relevant to generating the explanation. The 119 clinician’s judgment about the importance of information, it should be noted, is separable from the information that the patient considers important; the two parties could disagree greatly about this. The above three levels of interaction between clinician and patient engage the patient in a limited way. The clinician is typically the agent in charge of sorting out information relevant to the clinical explanation to be generated. The patient may have an active role in providing facts about her history or in asking questions, but the clinician is generally the agent determining how the conversation—and thus how the explanation— will go. Below, I argue for a different strategy. The Clinician-Patient Team as a Community of Knowers: The clinician-patient community exists prior to the many instantiations of doctor- patient interactions. An instantiation of the community is formed, for example, when a doctor and a patient meet in the exam room. An instantiation of the doctor-patient community is formed when clinical staffs of various specialties collaborate to “team manage” a patient’s cancer treatment. Another is formed when a social worker meets with a wife, siblings, and children to decide whether to continue life-support for an elderly parent. Yet before these occurrences, clinicians have had other encounters with patients. In medical school, future physicians are trained how to interact with patients in role-playing exercises. Patients, too, are trained how to be a patient: television shows like ER and plays like W;t give audiences insight on how they will be treated (both in morally forthright ways, and the possible abuses that may occur). Birthing classes teach expecting mothers how to navigate their labor in advance. Both the role of being a patient and of being a clinician is something we usually enter into with advance training and expectations. There are roles in addition to that of patient and clinician that come into play. Recently, the role of advocates—both patient advocates and research advocates—has developed as an important set of participants influencing patient care and biomedical 120 research. Patient advocates may be individuals who “keep watch out” for individuals in the hospital. Web communities may share discussions about their experiences of being ill or pregnant, sharing remedies, prevention strategies, and tips for navigating the medical system. Political groups, such as the women’s health movement or ACT-UP, try to enact policy changes at the institutional, state, or federal level.170 In this way, there are additional persons involved in treating illness, stretching beyond the doctor-patient pair. So, the team involved in caring for and explaining the problems of a patient can be the dyad of a single doctor and a single patient, or a more complex grouping, say, that of a clinical team, the patient, and the patient’s extended family.”' Note that for the interest of this paper, I will mainly discuss this doctor-patient community as instantiated by the relationship of a single doctor and a single patient. But again, I see their relationship as deriving from the larger doctor-patient community that comes prior. I also intend my comments to apply to larger, more complex, communities.172 I consider possible objections to this position below. At this point, one might object to calling the doctor-patient interaction a formation of a “new community” as I have done. For instance, some may argue that “community” implies too strong a connection, either of emotion, length of time, or other commonality, that is rarely there to unite patients and clinicians. These, though, are not the uniting factors of Nelson’s communities. Instead, her epistemic communities are united by shared principles and interactions on gathering and evaluating evidence. no Rebecca Dresser, When Science Offers Salvation: Patient Advocacy and Research Ethics (Oxford University Press, New York, 2001). m Here I am arguing that it is important to take seriously how families can be knowers. Hilde Lindemann Nelson and James Lindemann Nelson have argued for the moral importance of families as decision-making entities in The Patient in the Family: An Ethics of Medicine and Families (New York: Routledge 1995). m For example, the community generating explanations may also include drug companies, although such representatives are rarely in the examination room when the clinician and patient meet. Yet through their educational and advertising materials, drug companies do have an influence on how clinicians and patients understand their product, and the disease the product is marketed to treat. While I find this line of thought interesting and fruitful for additional questions, I will not take it up in any strong sense in this paper. 121 Others may argue that it is preferable to understand the doctor-patient interaction as closer to the collaboration of an interdisciplinary team. In such cases, people of different backgrounds and expertise come together to collaborate on a project, say, the work in epigenetics which may involve cbll biologists, geneticists, chemists, and computer scientists. I think little rides on this distinction. Nelson’s community or an interdisciplinary team come together for a given project, and often they pull from different backgrounds and resources. The members of an interdisciplinary team gain their various areas of expertise from other communities of knowledge; together they structure communal approaches to the project at hand. The new group or team that is formed need not be friends or close colleagues, since that is not how she intends “community” to be understood. These individuals come together to form new communities and sub- communities for specific projects, and often dissolve these communities as the project is completed. This happens in scientific projects, political projects, and, I argue, when patients and clinicians come together for certain projects, and dissolve their epistemic relationships after the projects are completed. As an analogy, consider the type of work done as part of a jury’s deliberation. A standard jury is made up of a collection of persons, each with their own backgrounds and expertise. Most jurors will not have significant legal expertise as part of their backgrounds. They come together to make epistemic judgments in the form of a legal verdict after hearing and evaluating the evidence of a case. They do engage with legal experts (e.g., lawyers who present evidence, judges who provide jury instructions to organize the rules for evaluation of the evidence). So, as legal laypersons, they must evaluate evidence according to legal standards (standards given to them by the judge of the trial as jury instructions), and make a legal decision based on the evidence. After the trial, the jury disbands. In this way, legal laypersons must come together and work as part of a complex legal mechanism. 122 Similarly, in generating medical explanations, clinicians and patients (each with their own various expertise) come together as part of an epistemic project that is not fully in either of their expertise domains. First, it is clear that clinicians and patients come together as parts of other prior communities. Second, the proj ect of generating a clinical explanation is not something that is derived exclusively from any one of those prior communities. There is collaboration—a merging of communities, backgrounds, knowledge, and assumptions—that allows both parties to generate the explanation. Some readers may still object that even in collaborative projects, two people cannot constitute a ‘community’ in a meaningful way. If scientific progress is interesting as a source of knowledge generation because of the epistemic mechanisms involved, e.g., peer review, evidence, experiments, then none of this is occun'ing in a pairing of two individuals. In other words, much of what is interesting about science cannot happen in this pairing. Despite this objection, I maintain that the doctor-patient relationship can constitute an interesting and meaningful community. Consider that a small chemistry lab may only have two research scientists involved. We still would describe them as part of the scientific community, and thus the research of these two individuals is likely meaningful. Although they, as two individuals, do not take part in all of the practices of science at any one time, they do engage with some (e.g., bench research), and may later take part in others (e.g., peer-review of data). Nelson would argue, and I agree, that the scientific pair are first part of a larger scientific community. They act as individual scientists only because of their prior training as scientists and inclusion in that community. More importantly, the community of scientists exists prior to these individual scientists. That is, a scientific community exists before this lab is established, and that community will also exist after this lab has completed its work and disbanded. Likewise, the collaborative community of doctors and patients in some way exists before the face- 123 to—face meeting of any give doctor and patient. Doctors are trained in how to encounter future patients; clinicians gather experience with patients. All of this becomes background to when a doctor meets with a patient. Similarly, patients have an understanding of doctors before meeting any specific doctor. Patients have cultural knowledge (say, from television and movie portrayals of doctor-patient encounters), and patients have past experiences with other clinicians. Thus, when a doctor and a patient come together to generate explanations, they are doing so not as two acontextual individuals. Their relationship has a preexisting structure and context. Thus, the doctor- patient community exists prior to that of any specific doctor and patient meeting, and this serves as the foundation for their development of explanations and other interactions. What is interesting for the sake of this project is that doctor-patient communities (and their instantiations of individual doctors, nurses, patients, etc.) must now generate knowledge claims, specifically, medical explanations. By conceiving of the group, not the individual, as the epistemic generator of the explanation, we avoid two common problems. The first problem involves an improved understanding of the goals of clinical explanations: if a patient fails to understand the explanation as generated by the patient- clinician team, then the explanation—not the patient—is faulty. The second problem avoided is the automatic move to prioritize clinicians’ contexts (over that of patients) as the appropriate location from which to generate an explanation. The First Problem The first problem involves an analysis of explanations that fail to relieve patients’ epistemic puzzlement. This is not the same as explanatory failure due to inadequate evidence or limitations on current scientific theories and understandings. Instead, we may ask the question: when a patient fails to understand an accurate medical explanation, 124 where should we locate the failure?173 Consider the standard strategies employed by Thagard’s explanation of ulcer formation. The explanatory inquisitor is the patient, who asks a why-question (Why did I develop an ulcer?), and the clinician provides an explanatory response (a CN I diagramed as a complex web of multifactorial interactions, or perhaps something closer to Railton’s ideal explanation).174 Standard accounts of medical explanation would hold that the explanation is accurate, and therefore it is not the source of the problem. The failure is located within the patient: his or her inability to grasp the complexities, the scientific language (or jargon), or other scientific theories involved in the explanation. These limitations might be correctable if, say, the patient could be trained in science, were the necessary time and resources available. But the explanatory failure is not due to the structure of the explanation. Yet these standard accounts of medical explanation leave much to be desired. If medical explanations are practical projects meant to improve patients’ understanding, then in this case the explanation has failed. Now, we could distinguish here between correct explanations and good explanations.175 This explanation is perhaps correct in that it reflects the way the world is (at least as well or better than other rival explanations). Yet it is not necessarily a good explanation since it fails to inform the patient, which was one of its main goals. This strategy would lead to various solutions, such as translating the explanation into terms the patient may understand, or educating the patient of the basics needed to understand the explanation. This solution, though, would need to be developed further, and remains relative to each patient. While I take it that this project could be developed, I see this strategy as only one of a set of possible strategies for ”3 I have argued in the previous chapter that such explanations fall roughly in the category of ontic explanations. Thagard’s CNI would be one such likely example, but not the only possible ontic structure. m Consider the CNI T hagard provides to explain the cause of an ulcer, diagramed in Chapter 3. ”5 See Salmon, Four Decades of Scientific Explanation, 148. Referencing Achinstein, Salmon notes that a correct explanation may not be a good explanation for any number of pragmatic features if it is “unsuitable to the knowledge and abilities of the listeners, or lacking in salience with respect to their interests.” Salmon generally does not worry about developing a theory of “good” explanations, in favor of exploring the logic of a correct explanation. 125 explanation. This strategy illustrates the idea that explanation involves a certain type of medical expertise; the translation and inclusion of patients as epistemic agents is not necessary. Instead, worries about patients’ understanding and their involvement are only important to the extent that the doctor’s beneficence worries about such matters. Instead, I favor an explanatory strategy in medicine that involves patients from the beginning, rather than as a secondary consideration. This comes about by viewing the agents of clinical explanation as a collaboration between clinicians and patients. An alternative strategy, one suggested in light of Nelson’s work, is to understand the doctor-patient community to be both the explanatory inquisitor and the explanation provider. The patient and clinician formulate the explanatory request. The why-question to be answered, then, is not determined exclusively by the clinician; it is generated by the collaborative efforts of the patient and clinician. Also, both the patient’s and clinician’s interests shape the explanatory response. This is in contrast to the standard approach, which privileges the explanatory interests of the clinician, and in large part considers the explanatory interests of the patient as secondary. This also implies that medical explanations are best considered as erotetic explanatory projects, rather than ontic projects. Salmon may be correct and ontic explanations are most frequently used by medical researchers. Yet as I have argued, ontic explanations often fail to meet patients’ explanatory needs. Instead, erotetic explanations emphasize the need to answer the questioner’s question (rather than follow a specific explanatory pattern, as is the case in ontic explanations). By considering both patients and clinicians as explanatory agents, erotetic medical explanations will need to meet a wider range of explanatory interests due to their different background knowledge, skills, and experiences. If we begin by understanding clinical explanations as generated by communities, not by individuals, we come to understand the clinician’s role in biomedical research explanations is much more complicated, and that generating clinical 126 explanations requires two-way (not one-way) interaction between clinicians and patients.176 One might also object to my strategy thus far by asking whether it is possible to develop an erotetic explanation that meets the explanatory needs of a patient, yet one which bypasses the idea of clinicians and patients as an epistemological community that I have argued for. Such an erotetic explanation would involve the clinician generating this on behalf of—or for the use of—patients. Such an erotetic approach would be an improvement over the ontic explanations of Thagard’s CNI. In this way, erotetic explanations may solve one of the problems from the previous chapter: specifically, it would allow medical explanations to be contextualized. Yet other difficulties remain, similar to those I noted with Thagard’s CNI. First, by keeping the clinician as the primary agent of the erotetic explanation, the patient remains disempowered as the subject of explanation. The clinician remains the only active epistemic agent. The clinician here seeks to inform the patient out of a sense of beneficence, but not out of a stronger epistemic duty. Second, this strategy ignores patients’ contributions to explanations. Patients often bring different information to be considered as part of the explanation. Sometimes, they will bring different ways of thinking to the problem, strategies that clinicians may not see (due to being habituated into other ways of thinking, or just lacking the proper position). Although I ultimately will argue for an erotetic explanatory strategy in detail in the next chapter, only a strategy that includes in a strong sense both clinicians and patients as epistemic agents will be useful to a theory of medical explanation. By recognizing clinicians and patients as an ”6 Again, it may be possible to develop an ontic model of medical explanations that includes the patient as an explanatory agent, and also considers the social context (and thus the power dynamics) of generating such an explanation. But I do not see the beginnings of such an explanation in the work of Thagard or Schaffner; other authors, such as Salmon, explicitly dismiss such a project as extra-explanatory (and thus uninteresting and messy). My original understanding of a good explanation, thus, are so different that I reject an ontic model of medical explanation for what I see a more promising pathway: developing an erotetic account. 127 epistemic community, we can view this community as the agent asking why-questions and generating explanatory responses. The Second Problem In the previous section, I argued that it is problematic that CNI often locates patients’ failure to understand as a problem with the explanatory response itself; instead, CNI often locates the source of the problem with the patient. I find this attitude dismissive of patients’ interests, and it is best avoided if possible. But there will at times be cases where the patient is confused or wrong about their explanatory requests. This leads to the second remaining problem: when, if ever, can a doctor appropriately reject patients’ explanatory requests? Thagard’s CNI does not address this question. We can assume that the clinician is sympathetic in including the patient in the explanatory process. Yet even in such collaborations some explanatory requests by patients will be poorly phrased or inappropriate to ask a clinician to provide a response. In the former case, the clinician may salvage the situation by editing or rephrasing the patients’ why- questions. The second situation, though, is more challenging. Even if the doctor and patient are attempting to work as a collaborative team, they may come from radically different contexts. In such cases, how can a shared context be navigated (or, how can they develop the shared context that will be the basis for this doctor-patient epistemic community)? And if disagreements continue, does the scientific context of the clinician have more weight than that of the patient’s non-scientific context in generating clinical explanations? Thagard’s CNI does not provide guidance for these problems. I will argue in the next chapter that a solution lies in understanding clinical explanations as erotetic explanations. I will argue for one possible solution using the work of Bas van Fraassen, which provides tools for clarifying the proper context for why-questions. 128 Conclusion In this chapter I have argued that the explanatory agents of medical explanations are best conceived as epistemic communities. By understanding epistemic communities, rather than solipsistic knowers, as the primary epistemic agent, Nelson has rightly emphasized that knowing (and thus explaining) is the work of groups, not individuals. Utilizing such a strategy for theorizing medical explanations benefits patients. First, it empowers patients by recognizing them as active epistemic agents, rather than as passive recipients of doctor-generated explanations. Recognizing the community of ‘patients’ as an explanatory agent provides theoretical space for patients as knowers and collaborators in medical explanations. Empowering patients as epistemic agents positions them to contribute to the generating and evaluating of explanations. Patients are able to contribute information and theories about their illness, in collaboration with clinicians. Hence, clinicians are not the only agents able to ‘know’ about patients and their illnesses. Second, it decentralizes blame for explanation failure due to patients’ inability to understand. It is not automatically the patient’s failure to understand the clinician’s explanation (as is implied by Thagard’s model, as well as by Salmon), but instead a flaw of explanatory structure. Finally, it highlights the theoretical and practical complexities of clinicians’ roles in generating medical explanations. Not all may find convincing my arguments that communities are the primary agents of explanation. Even if some fail to agree with my and Nelson’s conception of communities as primary epistemic agents, I have argued for other points that need clarification regarding the agents of explanation and their social situatedness. First, a good theory of medical explanations ought to provide a descriptive account of patient agency and participation, as well as provide a normative account for patients’ empowerment. Thagard’s proposal fails to take this into account; I have argued for one possible alternative approach. 129 Second, I have argued that medical explanations ought to be understood as contextual and personal, yet this need not devolve into relativistic explanations. The important aspect of explanation highlighted by the topic of medical explanations is that they are collaborative projects: both clinicians and patients generate and evaluate the explanation. Such a theory empowers patients as knowers and as active epistemic agents as a primary consideration, rather than as a secondary moral consideration. The contextuality of medical explanations, while raised here, has not been thoroughly articulated. This leads me to the question that will be the focus of the next chapter: how ought we structure the explanations that are derived from this collaboration between patients and clinicians? If collaborative communities both generate and evaluate the epistemic project of generating medical explanations, what should this process look like? I have argued against ontic explanations as the basis for medical explanations, since this structure too easily privileges the explanatory needs of clinicians and excludes patients’ participation. Instead, I argue in favor of an erotetic structure for medical explanations. For this, I turn to van Fraassen’s work on pragmatic explanations, which I will develop in the next chapter. The erotetic explanatory structure will allow two things. First, it better identifies how background knowledge and contexts will shape what counts as a good explanation, and it makes this issue legitimate and central to the explanatory project. Second, and related, it allows an improved strategy for dismissing improper explanatory requests, one that does not privilege the clinician as the stand-in for the scientific expert (and thus the only credible explanatory agent). Good requests for medical explanations are wider than those immediately answerable by clinicians, but not so wide as to allow for any and all explanatory requests. 130 CHAPTER 5: PRAGMATICS OF MEDICAL EXPLANATIONS Introduction In this chapter, I provide arguments for understanding “good” medical explanations as pragmatic or contextual, rather than as acontexual or ideal explanations. Pragmatic medical explanations, I will argue, avoid a number of the explanatory perils for clinician-patient explanations that arise when using acontexual explanatory strategies, such as Thagard’s explanations as causal network instantiations (CNI). What constitutes a “good” explanation under Thagard’s CN I derives from its completeness in identifying the causal web involved with a given disease. In this way, I compared Thagard’s CNI to Railton’s ideal explanations. In Chapter 3, I criticized as unhelpful the strategy of developing clinical explanations by “whittling down” ideal CNI explanations. Medical explanations fail to be good—that is, fail to reduce epistemic puzzlement—for a number of reasons other than incompleteness. For instance, poor explanations may fail to take into account the explanatory participants and what their needs are. These needs will not be universal, and thus must be considered contextually. Only after such questions have been answered can a “good” medical explanation be generated or a poor explanation improved. Standard theories of medical explanation have conceived of “good” explanations as those that maximize completeness of detailed descriptions of contributing factors and causal mechanisms.I77 While such explanations may meet the needs of biomedical researchers, such explanations are too cumbersome for clinicians to provide to patients. Yet clinical explanations should explain disease in terms patients can understand and utilize in their healthcare treatment decisions. A good clinical explanation is not a ”7 See my discussion of Thagard’s CNI in Chapter 3. 131 whittled down version of the real explanation. In generating clinical explanations, clinicians must consider other information that is important to patients, and also set aside information that may be important only to researchers. For instance, to what extent should clinicians consider what information the patient may need? If an explanation is correct in a technical sense, is its strength or weakness as an explanation determined by whether the patient is able to understand the explanation? How should you change the explanation to suit his or her ability to understand? How do you respond if a patient asks questions you cannot answer? What is the patient’s role in generating medical explanations? Standard conceptions of explanation in medicine do not provide sufficient tools for answering these questions. Mary Ann Cutter has argued that explanations of diseases are contextual even in research settings.I78 While this is an improvement over acontextual medical explanations, Cutter focuses on what I have called “research explanations”, those peer-peer explanations shared between clinicians and biomedical researchers.179 Cutter argues that value judgments influence research explanations, and such value judgments may vary between areas of biomedical research (e. g., virologists and public health researchers will make different decisions). For instance, value judgments will determine what level of evidence is necessary to accept a hypothesis or to decide what research projects are funded. While I find many of Cutter’s arguments to be right, they are of limited help. It may not be all that surprising to think that value judgments play a role in research explanations. This still leaves unanswered how we should understand clinical explanations, those generated by clinicians and patients (experts-lay persons) during ”3 Mary Ann G. Cutter, Reframing Disease Contextually, especially Chapter 2, “The Case of AIDS". See also “Explaining AIDS: A Case Study” in Eric T. Juengst and Barbara A. Koening (eds), The Meaning of AIDS: Implications of Medical Science, Clinical Practice, and Public Health Policy (New York: Praeger Publishers, 1989) ”9 While Cutter uses the term “clinical explanation”, she means something different by this than I do. Her term comes closest to my use of “biomedical research explanations." Her work on the contextuality of medical explanations is focused on this research level, rather than at doctor-patient conversations involving explanations of disease. 132 clinical encounters. Even if we accept that research explanations are generally contextual, neither Cutter nor Thagard provides tools to navigate the contextuality of clinical explanations. To remedy this, I develop a pragmatic approach to medical explanations, drawing from the work of Bas van Fraassen (The Scientific Image). Van Fraassen argues that dramatically different explanations can be given for the same event, and each could be equally accurate. Explanations for van Fraassen are necessarily contextual, as opposed to, say, the acontexual ideal-explanations described by Peter Railton.‘80 We can recognize that the explanatory goals of patients and clinicians may differ dramatically. A pragmatic response to disease allows that multiple explanations are not only logically possible but are necessary to meet differing needs of various inquirers. Thus, we must navigate the differences between what patients and clinicians need from a “good” explanation. Many miscommunications between clinicians and patients may be resolved with a discussion of the pragmatics of explanation. Often miscommunication results when the questioner and explainer are asking different why-questions, questions that will be answered by different explanations. While scientific explanations and medical explanations are connected, neither is reducible to the other. Applying van Fraassen’s work on scientific explanations, I propose that clinical explanations are best understood as pragmatic explanations. The ultimate goal is twofold. First, at the level of clinical practice, I provide clinicians suggestions for navigating the real-life difficulties of providing explanations of disease during clinical encounters. I show that a number of clinical miscommunications are due to bad explanatory strategies; a clear understanding of pragmatic explanations can correct many of these misunderstandings. '80 See Chapter 2 for my arguments on Railton’s ideal explanations. 133 Second, at the level of explanation theory, I apply van Fraassen’s theory of pragmatic explanations, focusing on the specific domain of medicine. This raises important points. First, I show that a pragmatic theory of explanation can clarify the complexly intertwined explanatory roles of clinicians and patients. Second, I show how the structure of van Fraassen‘s explanatory theory allows for clarification of explanatory requests, the occasional rejection of inappropriate explanatory requests, and the development of explanatory responses that inform patients’ healthcare decision-making process. This largely involves proper identification and use of contrast classes and relevance relations. In addition to van Fraassen’s pragmatics of explanation, I show that a common mistake in use of contrast classes can be remedied by instead using reference classes. My goal in applying van Fraassen’s explanatory theory to clinical explanations is to provide evidence for the utility of this theory. While I generally utilize van Fraassen’s theory, I make two important clarifications. The first involves his concept of background knowledge, and problems that arise when discussing clinical explanations. I address the problems that arise in discussing the proper background information when the explanatory agents have strongly different backgrounds, e. g., when doctors and patients as non-peers generate clinical explanations. I will also argue for a second clarification involving relevance relations: a relevance relation of social factors as causally related to disease must be considered for many clinical explanations. Such a relevance relation has not been identified in the examples van Fraassen uses, nor in the work of others discussing the pragmatics of explanation. 134 Sorting Through Questions: Distinguishing Between Interrogative Statements and Why-Questions Requests for medical explanations—like requests for all types of explanations— are often ambiguous. For example, the simple question, “Why is Jane sick?” can be interpreted in numerous ways. Consider the following possibilities: Why is Jane (rather than Mike) sick? Why is Jane sick (rather than being healthy, as was expected)? Why is Jane sick now (rather than being sick last week)? Why is Jane sick with breast cancer (rather than with ovarian cancer)? Some of these questions, once articulated, are more difficult to answer than others. There will be times when patients are unable to sort out which of these questions were intended. Other explanatory requests may actually be inappropriate to ask of clinicians, in that these requests fall outside medicine’s realm of explanatory ability, e. g., “Why did God allow Jane to become sick?” Sorting through this morass of intended and possible meanings is part of clinicians’ duties. Some may argue, though, that making such fme-grained distinctions is really the work of philosophers. Yet clinicians ought to have a general understanding of this landscape in order to be aware that such distinctions can improve communication with patients. Although this is often difficult and time-consuming work, I argue that such clarifications are important for improved doctor-patient communication. In the sections below, I provide distinctions and tools for making such work easier. An initial point of clarification about explanations of disease can be helpful here. We can distinguish between the patient’s interrogative sentence versus the intended meaning of the why-question. An interrogative sentence is the set of words that request a reply. This is the actual spoken or written question. The interrogative question, though, may remain ambiguous as to which idea(s) the sentence is asking. A concrete event can be described in different ways, depending on 135 which facts we cite and how we shape our description of the event. So, we can ask multiple, differing questions asked about the same concrete event. Often, even the interrogative question’s “topic”-—the event or item to be explained—is unclear. So, it is important to clarify the intended meaning of the question in order to understand how to construct an explanatory response that addresses the questioner’s intended topic. The why-question is the speaker’s intended meaning. When an interrogative sentence is uttered, the context determines which why-question is being asked. To take an extreme case, Julia and Jack can both ask the same interrogative sentence, “Why am I sick?”, but their personal context (in this case, who is speaking) changes the context in which the question is asked. In other examples, clinicians’ areas of expertise will shape the implied context. Say Julia is suffering from a gastric ulcer. Biomedical researchers may be more interested in describing how bacterial infection leads to ulcer formation. Julia’s G.P. may consider such an explanation unhelpful (since these bacteria are ubiquitous), and instead she may explain that some change, e.g., diet, stress, new medications, lead to the ulcer formation. This change of context thus alters the way in which we interpret the intended meaning of the why-question, which will in turn change which explanation we provide. The person seeking an explanation of a given phenomenon (e. g., sickness) will express her confusion through a given sentence. Because this sentence is in the form of a question, it is the interrogative statement. Before a response can be generated, the listener/explainer must determine exactly which why- question is being asked. The listener/explainer must rely upon both the interrogative sentence and some knowledge of the questioner’s context. Let us consider another example that (a) distinguishes between the ideas of an interrogative statement and a why-question, and (b) begins to show how the contextual background of the questioner changes the intended why-question. Willie Sutton, a convicted bank robber, was asked by a priest, “Why do you rob banks?” Sutton replied, “Because that’s where the money is.” In this example, the phrase “Why do you rob 136 banks?” is the interrogative sentence. In one sense, Sutton did answer the question; at least we can say he responded to the interrogative sentence and to one possible why- question. Yet it seems Sutton misunderstood the priest’s why-question, so his response was inappropriate and almost humorous. If the priest were asking why Sutton robbed the bank versus the hospital or library, then the answer “because that’s where the money is,” explains why Sutton chose the bank over other possible locations. But if the priest was intending to ask why Sutton robbed the bank versus some other action, like choosing not to rob banks, then Sutton’s answer does not make sense. Although Sutton responded to the priest’s interrogative statement, Sutton failed to understand the priest’s intended why- question. Sutton’s confusion is the result of a misunderstanding of the (intended) alternative possibilities!“ Similar ambiguities regarding the intended meaning of why-questions often arise in the request for clinical explanations, in which patients ask for explanations from clinical experts. In this dialogue, it will be important to understand what shapes the interrogative statements, as well as to identify the proper why-questions that are being asked. The interrogative sentence, “Why is Julia sick?” seems at first blush to be simple enough to understand; yet it can be interpreted in different ways. Why is Julia (rather than someone else) sick? Why is Julia sick, rather than healthy, the way she was yesterday? Clarifying the why-question is the fust step in determining what information will be helpful in answering the patients’ why-question. It should be noted here that Thagard’s proposed explanatory schema, medical explanations as causal network instantiations (CNI), utilizes why-questions, and this explanatory strategy can benefit from clarifying the interrogative statement. Although Thagard does not argue for this clarification, I believe it is not overly controversial, and that making such clarifications complements Thagard’s explanatory schema. For '8' Alan Garfinkel, Forms of Explanation: Rethinking the Questions in Social Theory (New Haven: Yale University Press, 1981), 21-22. 137 instance, Thagard admits that the ideal or complete CN I will be edited according to the question at hand, but he does not elaborate. Clarifying which why-question is being asked is a first step in determining which information from the ideal explanation will be appropriate to answer patients’ why-questions. While I have argued that such clarifications are beneficial, Thagard’s original schema (as well most other proposed medical explanatory schemas) does not provide means for making such distinctions. This oversight biases the case towards the idea that there is a single (or ideal) explanation for any why-question, and does not encourage us to explore further. Part of the goal for the next sections will be to explore how to navigate the “clues” or contexts, such that we can identify (a) the proper why-question being asked and (b) what types of information will best provide an explanatorily relevant response. To do this, we will look at an alternative to Thagard’s CNI for medical explanation, an alternative which draws from Bas van Fraassen’s pragmatic explanations as answers to why-questions. When this is addressed, we will note how van Fraassen’s pragmatics of explanations improves explanation activities in medicine. This explanatory strategy improves clarification of the why-question, allows identification of the proper context (either viewing it too broadly or too narrowly), for rejecting some explanatory requests as improper, and possibly editing these why-questions into something to which clinicians can respond. Thus, van Fraassen’s pragmatics of explanation has positive implications for medical explanations of disease, as well as doctor-patient communication. Van Fraassen’s Pragmatics of Explanation Van Fraassen claims explanations answer inquisitor’s why-questions. His explanatory theory is therefore an erotetic, rather than ontic, conception. The question (Q) takes the form of ‘why Pk?’, where Pk is the fact or phenomenon that is to be explained, that is to say, the explanandum. All why-questions have three components, 138 which can be described as an ordered triplet: , where: 1. Pk = topic 2. X = contrast class = {P1, P2, P3...Pk} 3. R = relevance relation The topic is the state of the world to be explained and is about what the why-question is asking. The contrast class is the set of alternative states of the world. The contrast class includes the topic, which is true and is about what the why-question asks. The relevance relation is a claim about which causal factors are relevant to the explanation. It is a sort of filter that determines which causal factors are explanatorily salient, and which will be set aside. To see how these components work, consider this brief example: Jane asks her son’s pediatrician, “Why are my son’s eyes blue, rather than brown like mine and my husband’s?” The topic of this explanation is “my son’s eyes are blue”. The contrast class involves possible eye colors {blue, brown}.182 The relevance relation appeals to Mendelian genetics involved in heritable traits like eye color. These three features (topic, contrast class, and relevance relation) must be properly identified to provide an answer/explanation to a why-question. These features of an explanation are structured such that the explanatory answer (A) will give reasons for why the topic (Pk) is true (or more explanatory), in contrast to the other members of the contrast class.'83 I will discuss each of these three features in greater detail below. A proper explanation will show that Jane’s child inherited certain recessive genes from each of his parents. In this combination, the phenotype was expressed as ‘blue eyes’, although other combinations were possible and those would have resulted in other eye colors. One of the implications of this model is that van Fraassen argues that dramatically different explanations can be given for the same event, and each could be equally '82 Had Jane instead asked, “Why are my son’s eyes blue rather than some other color?”, she would be asking a related but importantly different why-question. The contrast class would be expanded to include all possible eye colors {light blue, blue, blue-green, hazel, light brown, brown, dark brown}. Answering this would require a different explanation. ‘83 Van Fraassen, The Scientific Image, 143. 139 accurate. A common example involves a number of specialists who try to determine why a car crashed, which resulted in the death of the driver.184 An automotive engineer, a structural engineer in charge of highway design, and a medical coroner will each answer the question “what killed the driver?” in a different way. One expert might cite flaws in the car design; another might cite poor road conditions; the coroner might point to brain trauma as the cause of death. From the point of view of each of these experts, the explanations they give would each provide reasonable and accurate explanations. How can there be three (or possibly more) explanations for the death of the driver? Some might argue that these separate explanations are at best partial; in Railton’s terms, each person has provided part of the ideal explanatory text regarding the cause of the crash. What van Fraassen’s pragmatics of explanation provides is the important implication that each explanation can be correct, even if they disagree about the ultimate cause of the crash. In this way van Fraassen’s pragmatic explanations are complete explanations in that they should directly respond to the questioner’s contextual why-question.185 Rejecting the claim that an acontextual approach to explanation is even possible, van Fraassen instead argues that explanations are mutable given the context of the questioner. This is in contrast to rival explanatory theories that see explanations as independent entities to be discovered by the questioner. Each of the above experts is asking a different why-question about the same crash, and each why-question has been shaped by the context of the person asking the question. The mechanical engineer can ask, “why did the car take 40 feet to brake, rather than 20 feet (which was expected)? The civil engineer may ask, “why did the road conditions here (rather than at a different intersection) result in a crash? A coroner may ask, “why did the driver die in the crash, '84 Van Fraassen, The Scientific Image, 129. '85 Some may object that these are not “complete” in the sense of Railton’s ideal explanations. It may be that pragmatic explanations are only parts of the overall story we can tell, and thus similar to Railton’s explanatory texts. But I do not mean to beg the question here: I do not take it that van Fraassen believes that an ideal explanatory text (the complete story in a very strong sense) is ever told in science (or elsewhere), nor is it the ideal goal of explanation. We do not ask why-questions from such a context. 140 in rather than be injured but not killed, or even remain uninjured?” In the end, each of the different explanations answers a different why-question and is complete in its own way. Likewise, clinical explanations are necessarily contextual, where differences are based upon the interests of the questioner. Clinicians and patients can each seek an explanation of the same phenomena, but each party may require a different explanation. Van Fraassen’s pragmatics of explanation does not rank one of these as preferred or more accurate than the other, since he denies there is a single, acontextual explanation that can be provided.186 In the following sections, I discuss how generating clinical explanations by clinicians and patients can be improved by utilizing the features of van Fraassen’s pragmatic explanations. Topic The topic (Pk) of a why-question is the state of the world about which the explanation is meant to improve one’s understanding. In the car crash example above, all of the possible explanations focus upon answering the question, “Why was the driver killed?” The common feature to all their explanatory answers—‘the driver was killed’—— is the topic (Pk). The topic becomes a device to show what is common to the experts’ various perspectives. The experts’ why-questions about the crash share the same topic, even though they utilize different contrast classes. The topic of the why-question is the focus of the explanation. A good explanation shows that the topic is true (or at least more probable) in contrast to other states of the world. Contrast Class The contrast class (CC) is the set of possible states of the world; it includes the topic, which is true, and the other possible alternatives, which are false. A successful explanation will show why the topic is true or is “favored more than” the other members '86 Van Fraassen, The Scientific Image, 130. 141 of the contrast class. In this section, I examine examples of explanations that highlight the use of contrast classes. Then, I distinguish three types of work done by contrast classes that are useful in clarifying the task of giving medical explanations. As a means of clarifying that there is a range of possible meanings for patients’ why-questions, earlier in this chapter I distinguished an interrogative sentence from the why-question assumed by the questioner. We need to understand which of the possible questions is being asked so that we can provide a proper explanatory response. By clarifying the contrast class, we can identify which why-question is being asked. Recall Sutton’s mistake about the priest’s interrogative statement, in which he misinterprets the priest’s intended why-question. We can now further locate Sutton’s misunderstanding: Sutton’s error involves his use of what van Fraassen calls “contrast classes”, or what Garfinkle calls “altemative space”,187 or what Dretske calls “contrastive focus.”188 In order for an explanation to relieve the puzzlement of the questioner, the explainer first must identify correctly the why—question of the questioner; only then can the explainer generate an appropriate response for the questioner. Sutton mistook the contrast class of the priest’s question, “Why did you rob the bank?” as being a set of possible locations to rob: {rob the bank, rob the library, rob the hospital, rob a private residence}. The priest’s intended contrast class was instead a set of possible actions: {rob the bank, get a job and live an honest life}. Sutton’s failure to accurately identify the contrast class prevents him from responding to the priest’s intended why-question, and he, thus, fails to provide a good explanation. Let us consider another example provided by van Fraassen to clarify the distinction between interrogative sentences and why-questions. Consider the question, “Why did Adam eat the apple?””’9 One can ask this question with several different "‘7 Alan Garfinkel, Forms of Explanation. "‘3 Fred Dretske, “Contrastive Statements" Philosophical Review, 81 (1972) 411-437. I take these three concepts to be roughly equivalent for the purposes of this discussion. ‘89 Van Fraassen, The Scientific Image, 127. 142 meanings: I Why did Adam eat the apple, rather than someone else {Eve, the goat, the snake,. . . }? I Why did Adam eat the apple, rather eat than something else {the pear, the peach, the goat,. . . }? I Why did Adam eat the apple, rather than do something else with it {set the apple down, throw the apple at the snake, stare at the apple, ...}. Each of these possible why-questions would require a different strategy to answer. The first question is really about who ate the apple; the second question about what Adam ate; the third question is about what Adam did with the apple. Although each reading has the same form of interrogative sentence, “Why did Adam eat the apple?”, the intended why- question is different, and thus shapes what is considered an acceptable answer. Van Fraassen provides an example—a possible medical explanation involving a case of paresis, due to untreated syphilis—that further clarifies the importance of clarifying contrast classes. The ambiguity involved in asking unclarified interrogative statements is due to an uncertainty of the contrast classes involved. If a mother asks why her eldest son, a pillar of the community, mayor of his town, and best beloved of all her sons, has this dread disease [paresis], we answer: because he had latent untreated syphilis. But if that question is asked about this same person, immediately after a discussion of the fact that everyone in his country club has a history of untreated syphilis, there is no answer. The reason for the difference is that in the first case the contrast-class is the mother’s sons, and in the second, the members of the country club, contracting paresis.190 In this case of the mayor’s paresis, two different explanations are provided by changing the contrast class used. The first contrast class is about the people involved: {the mayor, his brothers (who are assumed to be syphilis-free)}. This difference would likely be explained by articulating how the mayor became infected (and/or how his brothers failed to become infected). The second contrast class is about the differences on why some '90 Van Fraassen, The Scientific Image, 128. 143 syphilis infections advance into paresis: {the mayor developed paresis, the other men with untreated syphilis have not developed paresis }. This second reading is more problematic to explain. We know that all these men had untreated latent syphilis. Yet no explanation may be available since we do not understand what causes some syphilis sufferers to develop paresis while others never do.'9' The contrast class allows for three different types of work to be done. Below, I describe these types of work, and how they benefit clinician-patient conversations. First, the contrast class allows us to clarify the intended why-question. Second, it allows a means for rejecting improper why-questions. Third, the choice of contrast class determines what information will be relevant for providing an answer. This will connect up to the relevance relation, as I will discuss. To Clarify the Question (and the Topic) The first benefit of the contrast class is that it allows us to understand what the questioner’s intended question is, thus decreasing ambiguity. By clarifying the members of the contrast class, we know to which set of alternatives the topic is being compared. In some cases, this lets the clinician differentiate between two different questions the patient may be asking. The clinician can then decide which why-question to answer. By identifying (or providing) the contrast class, the clinician clarifies the specific explanandum of the medical explanation. Without such clarification, there is a risk of providing incorrect explanations because the topic was chosen from an incorrectly identified contrast class. In some cases, patients may be unclear about their intended contrast class. Consider the mayor from van Fraassen’s above example. In a conversation with his doctor, this man may ask, “Why do I have paresis?” Such a question needs to be clarified before the '91 It is important to note, though, that just because we are unable to answer the why-question does not make it a bad or inappropriate why-question. It may be ultimately unanswerable, or perhaps it will one day be answered by updated scientific understandings of this disease. Either way, a good why-question (that is, a well structured explanatory request) does not necessitate that an explanation is available. 144 clinician can answer it, considering that the following three questions have different intended meanings due to their different contrast classes: Why do I (rather than my brother) have paresis? Why do I have paresis now, rather than last month or last year (since I have had the syphilis infection for a number of years)? Why am I infected with syphilis that developed into paresis, rather than being uninfected? After each contrast class is identified, each why-question will require a different explanatory response. Consider some of the questions we might ask regarding the health of Paul, a patient who has recently been diagnosed as HIV+; his partner, Nick, remains HIV-. If Paul asks, “Why am I HIV+?”, the topic of the explanation is “I am HIV+.” Paul may be interested in a range of issues about his disease. Hence we can identify separate explanation-seeking questions, each using different contrast classes. His single uttered question may thus generate a cluster of clinical explanations. The first we might explore utilizes the most basic contrast class {HIV+, not HIV+}. Here, the contrast class of {A, not A} is rather basic.192 But the contrast class can highlight other possible complexities. Paul could be asking a number of different questions that should be teased apart: why do I have this disease rather than another one; why do I feel ill today (rather than healthy like I did last week, or extremely sick like I did six months ago during my ‘bad phase’); why do I have this specific strain of HIV rather than another (say, the much hyped ‘super AIDS virus’); why am I HIV+, yet my long-term, monogamous partner (Nick) remains HIV-? Each of Paul’s possible why-questions utilizes a different contrast class, and thus, a clinician’s response will differ on each. '92 While this is a common example, I do not mean to argue that this is how all patients will begin an explanatory request. Consider, for example, that some patients will begin with a higher understanding of their disease state. Being thus informed, they may begin with more sophisticated contrast classes, e. g., why is my viral load per mL of blood {less than 20,000; between 20,000 and 200,000; greater than 200,000}? In other cases, a clinician may eventually become a patient, who herself begins the explanatory process with some sort of contrast class regarding, say, her cancer formation. For this clinician, the contrast class may be more sophisticated than that of a medical lay person, but may not be as sophisticated as that of an oncologist (a specialist). 145 ‘e Notice that the question “Why is Paul sick, but Nick is not?” may be answered wholly or partially by an explanation of Nick’s condition, rather than about Paul’s condition. That is, we can edit the question to ask only about Nick’s condition: Why is he still HIV-, given that we have strong reason to suspect that he could have become infected through his sexual contact with Paul. There are different types of explanatory routes available that need further sorting. Some individuals appear to be protected from developing AIDS or even from becoming HIV infected, but for different reasons. Some genetic variations allow patients to become infected with HIV, yet seemingly not develop AIDS over a period of many years. Other persons develop immunity to the specific HIV strain of their partner, an immunity that does not protect them from other HIV strains. So, in asking about Nick’s health, we might use the contrast class {no detectable immunological response; HIV infection occurred; localized HIV immunity developed}. In this way, we now have a more interesting set of possibilities than the basic contrast class of Nick is {HIV+, HIV -}. One of the important points of this example is that it shows the complexity of generating why-questions and explanatory responses in clinical settings. As will often occur in a conversation between a doctor and a patient, the patient may be at first unclear which why-question to ask. But providing an explanation may not end the conversation. One explanatory request leads to another, and this is often how clinical conversations occur. Different contrast classes (and thus different explanatory responses) are considered and discussed in the back-and-forth between clinician and patient. It is possible the patient, given his state of being shocked or overwhelmed by his diagnosis, is unclear as to which contrast class he intends. In such a case, the explanation the clinician ought to provide is not obvious. Remember that this work is done as part of a back-and-forth dialogue between the clinician and patient. The clinician may need to guess at how to fill in the contrast class. If it is discovered this was not what the patient intended, they can backtrack and recreate an improved contrast class. 146 Allows for Rejection of Improper Questions A second type of work done by contrast classes is that they determine whether why-questions are answerable in a meaningful way by scientific or medical explanations. As in the earlier paresis example, the first formulation of the question (Why did he develop paresis?) is answerable by making reference to his previous untreated syphilis infection. Yet other why-questions may be unanswerable for various reasons. If we ask why the mayor developed paresis (while the other members of the country club only have untreated syphilis), this is a well-formed question that cannot be explained with today’s i scientific data. ’93 Some why-questions are unanswerable by current scientific information, even if they have proper CC. Although an explanation is unavailable at the moment, there probably is some fact of the matter, and further medical research may uncover this truth. Some topics may also be fundamentally indeterminate or random, and a declaration of such may be the best science can do. Another variation might be that some explanations are possible, but they would require such a wide and detailed set of information that it may be nearly impossible to provide such an explanation. This may require collecting information that will take centuries to collect, and thus impossible for current science. In some similar situations, it may be possible to collect such vast amounts of information, yet this remains a highly impractical task. In these latter cases, a fact of the matter may exist, but it may be impossible or too impractical to provide an explanatory response. '94 In such cases, the why-question and contrast class may be formed correctly, but other limitations prevent an explanation from being crafted. '93 Note that van Fraassen says it cannot be explained, yet I think this may overstate the case. We cannot today explain the difference, but it is possible there could be the future discovery of the mechanism that makes the difference between syphilis and paresis. '94 Here, I am thinking of the something akin to Railton’s “ideal explanatory text” as an explanatory response. The limitations of these are discussed in greater detail in Chapter 2. In brief, it is questionable as to whether science ever is able to generate such complete explanations. Even if such complete, ideal 147 Explanatory requests can be unanswerable for reasons other than a lack of information or limitations in current scientific understanding. Some explanatory requests for medical explanation may be unanswerable due to utilization of improper contrast classes.I95 Consider the most basic case of a bad contrast class: the topic of the why- question is false. A false topic contradicts van Fraassen’s explanatory requirement that the topic be true. Why is this important? If it is not the case that the mayor has paresis, it makes no sense to ask, “Why does he have paresis?”, no matter what contrast class is used. So, why-questions with false topics are rejected because of the topic choice, not because of choice of contrast class. Let us consider a second reason to reject a why-question given the contrast class: the topic of the why-question is true, but an incorrect contrast class is used. If the mayor were to ask, “Why did my syphilis infection develop into paresis (rather than tuberculosis)?”, we can reject his question as having an improper CC. The question is bad in that TB is not a reasonable contrast event to developing paresis. The explanation of TB does not involve syphilis; whether or not the mayor ever develops paresis has no causal relationship to whether he also develops TB. There are other ways in which contrast classes can be wrongly—or, unhelpfully— chosen. Consider the earlier exarnple of Paul and Nick, who are HIV + and HIV -, respectively. If Paul asks, “Why am I sick, but Nick isn’t?”, this is a case of a poorly constructed contrast class. (Written a different way, one might ask, “Why is it the case that { Nick, Paul} is HIV+?”) While an explanation may be generated such that it picks out the topic as true (it is the case that Paul is HIV+, and Nick is not), there is something unfair or rather ad hoc about this contrast class. The contrast class, as originally explanations are available, it is not clear that these are beneficial to clinical explanations that I am investigating. '95 For more on this point, see “Van Fraassen on Explanation”, Philip Kitcher and Wesley Salmon, The Journal of Philosophy (1987): 3 l5-330 (see esp. 315). The authors are strongly critical of van Fraassen’s explanatory model given certain weaknesses they identify, mostly regarding van Fraassen’s inability to solve problems of explanatory asymmetry. Still, they complement this model’s strength in that it provides the mechanism for identifying and rejecting inappropriate explanatory requests. 148 ’l developed, wrongly assumes a certain state of the world before generating the explanation: that is, that only one of these two men could be HIV+. It could be the case that both Paul and Nick are HIV+. The original question, though, can be salvaged in two different ways. First, we could ask two different why-questions: “Why is Nick HIV+ (rather than not)?” and “Why is Paul HIV- (rather than not)?” By separating the why-questions, we can generate accurate explanations without the confusion noted above. A different line of thought for improving the question comes from Fred Gifford, who uses Wesley Salmon’s concept of “reference class.” A reference class is a population with a given set of traits or characteristics. Gifford might argue that Paul’s original question, “Why am I sick, but Nick is not?” is better understood through using reference classes than through using contrast classes. Contrast classes are a set of effects, outcomes, or phenotypes, a set that includes the topic of the why-questions. Reference classes instead focuses on the causal properties shared by a group or population, e. g., the members of the group of people with blue eyes, or the group of people who are HIV+.l96 In this way, utilizing reference class is about asking which population (with a set of shared characteristics) is salient for generating an explanatory response, rather than which property is identified by the contrast class as salient for an explanatory response. We can ask, what makes A different from the other members of that population. In this way, the use of reference class makes sense as a useful tool for pragmatic explanations in medicine, even if it is not a concept taken up originally by van Fraassen. By editing Nick’s question to utilize reference classes, we can instead ask, “Why is it the case that Nick (who is a member of the population of individuals with a high exposure risk to becoming HIV infected) remains uninfected?” A good explanation then rests on identifying what further property is needed for the result to occur (i.e., Nick ‘96 Gifford, “Explanations of Difference” (unpublished manuscript). 149 becoming infected) besides all the properties shared in the population (e. g., high risk of exposure). Relevance Relation How are certain conditions or information determined to be relevant to an explanation, while others are viewed to be irrelevant? Van Fraassen gives us his condition of the “relevance relation” (R), but he unfortunately does not develop this criterion in detail. Relevance relations are a sort of filter that determines which types of factors to cite as explanatory—that is, which causal factors you are interested in—within a certain context. Related to one’s explanatory interests, the relevance relation should specify the way in which the explanatory factors show that the topic is true and the other members of the contrast class are false (or at least less probable). As with the contrast class, the relevance relation is “determined by the context” in which the why-question was asked. Little more is said about the nature of the relevance relation, but we can see how it functions in some different examples. Through his examples, van Fraassen addresses certain types of common relevance relations, ones that are common in the philosophy of science literature: “physical necessitation, being etiologically relevant, fulfilling a function, statistical relevance, and, in the fable of ‘The Tower and the Shadow,’ a relation of intentional relevance.”197 While there may be multiple causes of any event, our choice of what we pick out as “the cause” is highly determined by our context and our explanatory interests. Recall the earlier car crash example; each inquisitor may explain the crash by making reference to a different relevance relation: 197 Kitcher and Salmon, “Van Fraassen on Explanation” p 325. This latter relevance relation, intentional relevance, has been highly criticized by Salmon and Kitcher. In my work, I do not defend this relevance relation, but merely point out that van Fraassen considers its use possible and appropriate in some situations. 150 Consider how the cause of death might have been set out by a physician as ‘multiple haemorrhage’, by the barrister as ‘negligence on the part of the driver’, by a carriage-builder as ‘a defect in the brakeblock construction’, by a civic planner as ‘the presence of tall shrubbery at that turning’.”8 In the earlier example of the question to Willie Sutton about why did he rob banks, the priest intended a relevance relation of intentional relevance (what motivated his actions). If we ask why does a copper strip expand when heated, the relevance relation is one of causal necessity about how metals respond to temperature changes. In a successful explanation, the relevance relation determines the kind of factors to be used to pick out (or favor) the topic from the contrast class (the list of alternative answers), makes the topic more likely, and the other possibilities less likely. As van Fraassen states, “A proposition is called relevant to Q exactly if A bears relation R to couple .”199 The relevance relation does certain distinct types of work. A relevant answer will pick the topic out of the contrast class and show that it is true: if we ask, why is Jane sick, a relevant explanation will show why it is the case that Jane is sick, as opposed to Jane being healthy. Van Fraassen purposefully does not set limits on what counts as a “good” or “genuine” relevance relation. He argues that different explanatory contexts will require different relevance relations. For the purpose of better understanding medical explanations, though, one can ask what relevance relations are best suited to medical explanations (as a specialized subset of all possible explanations). The types of relevance relations that van Fraassen utilizes in his examples can be labeled (roughly) as being physically necessary, etiologically relevant, fulfilling a function, statistically relevant, and a relation of intentional relevance. Below, I will look at how some of these are commonly used in medical explanations, and add to this list of relevance relations. '98 N.R. Hanson, cited in van Fraassen, The Scientific Image, 125. '99 Van Fraassen, The Scientific Image, p 143. 151 The function of the relevance relation is to determine what shall count as a possible explanatory factor. As Gifford notes, relevance relations organize the explanans.200 They work as a filter for sorting out which factors we will cite as explanatory (and which we will set aside as unexplanatory or uninteresting) in our attempt to explain why P in contrast to Q is true. The relevance relation may sort the type of explanatory account that is being requested—say, a functional, a causal-mechanistic, or probabilistic explanation. The relevance relation may also shape how far back in a causal chain the explanation should go, given our explanatory interests.201 The relevance relation, much like the CC, relies upon the interests and goals of the questioning person. Recall that in the car crash example different experts, because of their differing interests in the crash, will provide different but equally accurate descriptions of why the car crashed. The automotive engineer, for example, will pull from a different set of possible alternatives (a different contrast class) than the set from which the civil engineer pulls. Their interests determine what information types are relevant; this is a choice of relevance relations. The automotive engineer will look at factors of the crash such as braking distance, momentum, and the mass of the vehicle. The civil engineer will instead look at the road’s factors, such as its slant, curve, and the driver’s visibility at the intersection of the crash. In this way, the CC and R are complexly intertwined, but more can be said. The choice of R will often influence the CC, and vice-versa. This development of the relevance relation is where van Fraassen differs dramatically from other theorists of explanations. Just as the contrast class was one 20° Gifford, “Explanations of Difference.” 201 Richard C. Lewontin argues that we often mistake what we mean as “the cause” because we have not gone far enough back along the causal chain. The example he uses is that we often speak of asbestos fibers or harmful chemicals as the cause of workers’ health problems. Instead he argues the cause is better identified as the social forces of capitalism. If asbestos had not been invented, likely some other factor would have caused workers’ illnesses. Yet the more salient cause of disease—capitalism’s virtue of maximization of profit (which is in contrast to maximization of worker safety)—would still remain (Biology as Ideology: The Doctrine of DNA (New York: Harper Perennial, 1993): 45. 152 fi possible tool to sort out misunderstandings, the relevance relation also allows a theory of the pragmatics of explanation to account for differences between explanations due to the context of the questioner. This should allow for explanations to be improved and be more helpful to some cases. It allows explanations to utilize different levels or different types of information, rather than assume only one sort of information is correct. Recall van Fraassen’s comment, “to ask that their explanations be scientific is only to demand that they rely on scientific theories and experimentation, not on old wives tales.” 202 As I argued in earlier chapters, some why-questions may at first seem like requests for medical explanations, but they are really asking for explanations based on religious faith, the law, or some other (non-scientific) discipline. While these may be coherent why-questions with some type of explanation possible, these are explanations outside the domain of medicine. Thus, identification of relevance relations will sometimes show why some explanatory requests are inappropriate explanations to ask from clinicians qua clinicians.203 A common relevance relation used in medicine is that of etiological relevance.204 In the earlier example of the mayor, his having contracted syphilis is etiologically relevant to answering the question of why he has developed paresis. HIV infection is etiologically relevant to developing AIDS. In explaining why these patients develop paresis or AIDS, clinicians will typically draw from the scientific components of medicine that describe the infection and progression of these diseases, including process 202 Van Fraassen, ‘Ihe Scientific Image, 129 20’ Again, identifying this problem does not require the end of the clinician-patient conversation. In such a case, the clinician can identify this contrast class, tell the patient that he/she cannot answer such a question. But then, the clinician can continue the conversation, asking the patient whether a different contrast class would be helpful, a contrast class that ultimately leads to an explanation the clinician can provide successfully. ’04 It may be argued that ‘etiological relevance’ is a more narrowed version of ‘causal-mechanistic relevance’. I find this likely, but much of the literature separates the two, rather than seeing one as a subset of the other. It may be possible that some statistical and deeply indeterminate information is used as part of etiological explanations, and thus they are not purely causal«mechanistic. Or, it may be a relic of the domains producing different categories of explanations, e.g., physics and medicine. 153 of infection, viral activities at the microscopic level, or general theories of disease processes.205 Notice that ‘etiological relevance’ really refers to causal patterns that are clinically useful and that have become accepted as generalized patterns of disease (rather than about the experience of any one individual). As Robert A. Aronowitz writes on how these patterns of medical investigation and diagnosis became common: It might be argued that with the ascendancy of the germ theory of disease in the late nineteenth century. . ..it was discovered that particular microorganisms caused distinctive pathological derangements and clinical presentations, this etiology became the prototype for explaining most diseases and sickness in general, up to and including the credo of contemporary molecular biology: one gene, one protein, one disease. Individual factors such as the role of emotions, lifestyle, and social class in the etiology, appearance, course, and distribution of disease were, in the course of the twentieth century, relegated to the margins of medical and lay concerns.206 ‘Etiological relevance’ becomes shorthand for certain patterns of causal information. Some causal information is given explanatory priority (e. g., genetic, microbial activity, diet), while others were seen as minimally explanatory, if they were considered at all (e. g., emotion, socio-economic status). While these latter factors could be explanatory, clinicians became unaccustomed to using such information in clinical practice. Hence clinicians often presuppose a certain relevance relation, while overlooking other possible relevance relations. Etiological relevance also prioritizes proximal causes, rather than following the causal pathway further back. For example, Rose had seen her doctor numerous times over the past year for respiratory difficulties. Many of these were diagnosed as lung infections, and were treated with antibiotics and other drugs to improve her breathing. Her clinicians had 205 These two diseases are interesting in that the progression from HIV infection to full-blown AIDS is relatively well understood. This is not the case, though, regarding the progression of syphilis into paresis. 20" Robert A. Aronowitz, Making Sense of Illness: Science, Society, and Disease (New York: Cambridge University Press, 1998) 8. 154 become habituated into seeing Rose as having a reoccurring problem, and continued to prescribe treatments to ease her symptoms and treat the infection. This treatment seemed to work, but only for two or three months. On her last visit, though, Rose saw a different physician. Noting the history of the same symptoms, the clinician asked Rose about her daily life, her family, and her job. The clinician learned that Rose, a janitor, was exposed numerous times each day to harsh cleaning chemicals. By inhaling these chemicals, Rose was irritating the tissues of her throat and lungs, which in turn lead to the infections. The clinician repeated the past prescriptions, but with one addition. The doctor asked Rose to wear a respirator while working with the cleaning chemicals.207 The point here is that etiological relevance can habituate clinicians into certain patterns of drinking. Often, such patterns of thinking are beneficial. But in other circumstances, it is beneficial instead for clinicians to rethink the case if they explicitly try to understand the situation in different terms. In some cases, this means employing different types of information as explanatory, and in other cases it means delving further back along the causal processes for a different explanation of the situation. How should this inform our discussion of relevance relations in clinical explanations? Relevance relations work to include (or exclude) different types of information as important to the explanation. This decision should be made carefully and explicitly, so to best respond to patients’ and clinicians’ why-questions. If we rely upon standard relevance relations like the narrow version of ‘etiological relevance’, then we restrict what information we consider. This narrowing can, in turn, curtail treatment and prevention options. For instance, while HIV can be explained by its viral activity or the genetics of the virus, a public health worker may be more interested in the social forces involved that cause some populations to be at higher risk of infection that others.208 The 207 This example was in part inspired by a conversation with David J. Doukas, MD. 208 Of course, different viewpoints will instigate different explanations of HIV/AIDS (and thus utilize different relevance relations). It may be possible to develop one using statistical information, or a failure of bodily/ immunological function, etc. For an overview of different etiological, causal, statistical, and other explanations of HIV/AIDS, see Mary AnnG. Cutter’s “Explaining AIDS: A Case Study” in The Meaning of 155 patient may be more interested in transmission of HIV, either to prevent a future infection or to understand the circumstances of how they have already become infected. So explanations given by physicians will often be disappointingly limited for patients. By identifying the relevance relation explicitly, the patient and the doctor have means for articulating which causal factors they consider to be explanatorily salient. This illumination allows for disagreements and misunderstandings to be rectified more easily. Thagard’s CN I model of explanation does not provide a filtering resource like van Fraassen’s relevance relation. This omission may wrongly give clinicians the belief that they are utilizing complete or ideal explanations. Yet in real practice, given the limited resources of time and information, they are often making decisions (sometimes consciously, sometimes unaware) about how to narrow down the information they consider as part of explanatory responses to patients. By making explicit van Fraassen’s resource of the relevance relation, clinicians can become more self-aware, self-critical, and investigate new approaches to clinical conversations with patients. This also allows patients the means to understand how and why explanations of their illnesses are being generated in a certain way. By articulating relevance relations as part of doctor-patient conversations, patients are better able to edit the conversation to meet their own explanatory interests. That is, at certain points, patients may say, “Well, I understand that...but what I’m really interested is this. . .” Given that most diseases have a number of etiologically relevant factors, clarifying the relevance relation often will mean drawing from some medically relevant information, while holding other information as set or uninteresting. Consider the disease phenylketonuria (PKU).209 If we assume that the genetics of the disease are held stable (given that we have no genetic therapy currently available), then the disease is often AIDS. In this article, Cutter means by her phrase “clinical explanation” something much closer to what I call “research explanations”—those peer-peer explanations generated by and shared between clinicians. 20” Phenylketonuria is a genetic disorder that is characterized by an inability of the body to utilize the essential amino acid, phenylalanine. 156 explained as caused by dietary factors (and treated through changes in diet). If the dietary norms of a population are taken as set or standard, then the differences between people with or without PKU are best explained according to genetic variation.210 In this way, the relevance relation acts to filter out either genetic information or dietary information as irrelevant, given the context from which we examine PKU. Relevance relations may also determine what level of detail is appropriate to an explanation. This determination is not about what type of information is helpful (e. g., functional versus statistical information), but about instead is about how much detail or at what level this information should be portrayed. First, other relevance relations may determine how much detail about a causal pathway is relevant to the explanatory context, or how proximal or distal an explanation is requested. Consider a patient who wants to understand why she developed a cancerous tumor. A more proximal explanation will begin with cell growth that remains unchecked by the body’s immune system. A more distal explanation may involve the environmental factors that contributed to the original cell mutation, the inherited features that minimized her body’s ability to repair this environmental damage, etc. Generally, an explanation is good if it gives the right amount of detail. Explanations are bad if they do not give patients enough information to understand their situation (thus preventing informed decision-making). Explanations are also bad if they provide too much information, thus confusing the situation. Second, relevance relations determine at what level the explanation is set. As Hilary Putnam writes about the relative goodness or badness of explaining why a one- inch square peg will not fit into a one-inch round hole. The simple explanation is that the geometry of the situation prevents this, and likely this is a sufficient explanation. It is possible, though, to give a highly complicated explanation at the atomic level describing atomic positions and force, and then why despite multiple trajectories the peg fails to pass 2'0 Gifford, “Explanations of Difference” and Gifford, “Genetic Traits”, Biology and Philosophy, 5 (I990): 327-347. 157 through the hole. Yet such an explanation, while technically accurate, is too detailed, too dreadfully complicated to be grasped easily?” I would like to consider an objection. Those who are explanatory objectivists (those who believe explanations are acontextual) will object to my position, saying that while these are features of a good explanation, they are not necessarily features of an explanation proper. Salmon and Kitcher have criticized van Fraassen’s explanatory strategy, especially his refusal to set limitations on the relevance relation. They argue this avoids articulating an “authentic” relevance relation, and thus leaves open the possibility that any and all information can ultimately become explanatorily relevant.212 It is not my goal in this paper to make clear a conclusion to that particular argument. Instead, I will point to common relevance relations often used in medicine. There will not be any single relevance relation appropriate to every clinical explanation. Most commonly, etiological relevance will be used, i.e., bacterial, viral, or genetic etiology. Such relevance relations are included in van Fraassen’s original work. Beyond this, though, I have argued that a relevance relation that includes social forces will also be necessary for clinical explanations. Such a relevance relation, I argue, is legitimate in generating and evaluating medical explanations. The resistance to using such relevance relations often has more to do with sociological aspects of clinical practice and medical education, rather than a deficiency in such explanations.213 2" Hempel, Aspects of Scientific Explanation, 42-43. 2'2 Kitcher and Salmon do offer one possible solution, though. They suggest a more modest project, of determining relevance relations for a particular science (they suggest physics) in a given time period. Then, genuine relevance relations can be separated from those that are not. I take it my project has articulated the difficulties, though, in articulating the “genuine” and “false” relevance relations in medicine (Kitcher and Salmon, “Van Fraassen on Explanation” 326). 2” There remains a question about the relationship between R and CC. Does the choice of CC shape the relevance relation, or vice versa? For the most part, van Fraassen remains unclear on how dynamic this relationship is. In his paper “Explanations of Difference”, Fred Gifford argues that it is often a mistaken strategy to consider that a choice of RR structures the CC, but the choice of CC often structures which relevance relation we utilize. For the purpose of this paper, I will not take a stand in this interesting debate either way. It is enough for my purposes to identify how either feature (R or CC) is a location where explanatory misunderstands may occur. 158 Background Knowledge The final facet of a pragmatic explanation that van Fraassen identifies is the role of ‘background knowledge’ (K). Background knowledge is the information needed to answer why-questions, information that is not contained in the CC or R but is needed in order to pick out the topic—that is, show that the topic is (more probably) true. Recall that van Fraassen claims that all explanations are structurally similar. What makes scientific explanations unique is that they draw from scientific background information and theories. Because of the role of background information, medical explanations are an interesting example for consideration for two reasons. First, while medical explanations use scientific information, they are not scientific explanations. As I have argued previously, both scientific and non-scientific considerations come into play in generating clinical explanations.” The relationship between medicine and science is important, but not all of medicine is reducible to science.215 Even when focusing on the biomedical aspects of clinicians’ expertise, do all medical disciplines and sub-disciplines share a common background knowledge? Second, the agents of clinical explanations—clinicians and patients—at times will have radically different background knowledges, often differing greatly on their level of scientific expertise. Is there a “genuine” K for clinical explanations shared between clinicians and patients? While I take it van Fraassen would not deny these complexities, he does not address them in The Scientific Image. His discussion of scientific explanations assumes scientists generating explanations with and for other scientists. Yet this model does not smoothly translate to the generating of clinical explanations. Let us consider what background knowledge clinicians utilize in generating medical explanations. One difference in K involved in medical explanations—as opposed 2" See especially Chapters 1 and 3. 2" Munson, “Why Medicine Cannot Be a Science”. 159 to the scientific explanations van Fraassen typically considered—involves the background knowledge shared between those persons asking for and those generating explanations. In scientific explanations, scientists can be said to share roughly a general understanding of current scientific knowledge and practices, which can be taken as their shared background knowledge. In such cases, it seems legitimate to suggest that scientists may give partial or gappy explanations, or what Hempel calls explanatory sketches.216 Some information is assumed known already by (or readily available to) other scientists. This partial explanation is only a part of a larger, complete explanation, which could be developed in detail if necessary. This strategy, though, is unhelpful for two different reasons. First, it necessarily assumes that a single, complete explanation is possible. I have argued that this conception of what an ideal explanation is leads to problems for clinical explanations.217 Second, it describes background knowledge as static and shared between all scientific domains. The validity of this needs to be examined rather than assumed. Is it reasonable to assume that biologists and physicists utilize the same background knowledge when generating scientific explanations? We can see quickly that this may not hold, given how dramatic differences in knowledge, skills, and practices between scientific disciplines and specialties. Kitcher and Salmon raise this concern, but ultimately find it unproblematic: Since, typically the person who asks the question Q might be someone with a different body of knowledge from the resgondent 5,, we might be tempted to say that two different contexts are involved.2 2’6 “Gappy” here is Mackie’s term. See Schaffner, Discovery and Explanation in Biology and Medicine, p 301—302. 2” See Chapter 3. 2’8 Kitcher and Salmon, “Van Fraassen on Explanation” 318. 160 For instance, the background knowledge in physics is arguably quite different from that of biology. Kitcher and Salmon continue, though: It seems more in keeping with van Fraassen’s approach, however, to understand that 5,, and S, are operating in a common context with a common body of backgrgund knowledge K determined roughly by the state of science at the time. So, a solution for uniting seemingly disparate background knowledges of various scientific sub—disciplines is to understand K as something other than the knowledge held by any individual (which, as mentioned earlier, may be “gappy” knowledge). Instead, I would agree with Kitcher and Salmon that K is the set of current scientific knowledge shared throughout the scientific disciplines. This understanding of K allows for a broader understanding of the shared knowledge that is used to generate scientific explanations, even between different sub-disciplines, as well as peer-peer medical explanations (those shared between biomedical researchers).220 Consider the case of a surgeon and an immunologist trying to understand the case of a given patient. Because of their different trainings, these clinicians have different backgrounds and different skills as experts. Even though both may be trained as doctors, the neurologist may not understand the subtleties of the case as the immunologist does, and vice versa.221 We can say, though, that the general background knowledge (K) shared by these clinicians is the current state of medical knowledge as reflected by the medical community’s consensus. 2‘9 Kitcher and Salmon, “Van Fraassen on Explanation” 3 l 8. 22° While I do not make them explicitly here, there are interesting connections here to be made with Lynn Hankinson Nelson’s understanding of knowledge as generated by epistemological communities first, rather than by individuals. For more on Nelson’s work, see my Chapter 4. 22‘ Recall that in the car crash example, the various experts involved each were able to generate accurate explanations, but this did not require that they each understood the other alternate explanations generated by the other experts. 161 This move, though, to unite background knowledges between different biomedical communities cannot work for expert-layperson medical explanations. Patients typically will not share clinicians’ scientific background. Should the scientific-K automatically trump the lay-person-K? Does the patient raise the question, but the clinician answer it according to her/his background knowledge only? While this may be one solution, there are other preferable alternatives. Here, my arguments from the previous chapter (Chapter 4) may be helpful. I have argued that in clinical explanations the epistemic questioner is best understood to be more than one person, that is, the epistemic community comprised by patients and clinicians. It may be that if the clinician-patient team comes together to shape the why-question through conversations, they also similarly may construct the proper set of background knowledge needed to answer this why-question. This new background knowledge may include scientific knowledge, which is brought to the table by the clinician. But the patient, too, brings various sorts of knowledge: knowledge of her own experiences, of social knowledge.222 One might question this move, asking clinicians to step outside their normally understood range of expertise. To ask the clinician to be responsible for knowledge of social forces may seem a problematic move: this is to ask for an explanation using information the clinician may not be comfortable with. First, this challenges the view that they (clinicians) are the experts regarding medical matters. Second, it asks clinicians to make judgments about matters outside the immediate concerns of their own patients by adding the responsibility of considering social groups as well. 222 There are likely many ways in which the epistemological communities generating clinical explanations expand background knowledge (rather than seeing it as static). Patients may cause clinicians to expand K. Remember the experiences and understanding of racism that Uncle Bill brings to the table, information that is outside the clinician’s expertise and familiar understanding. Clinicians of different backgrounds, when working together, may expand the K used. Consider doctors with different background, i.e., osteopathic versus allopathic (perhaps versus herbalistic/holistic). These are different backgrounds in medical training, they speak different languages, they identify different problems, and they approach treatment differently. All of these backgrounds are grounded in something like “current conceptions of science”. 162 Yet if I am right in arguing that patients are epistemic agents of medical explanations, they often ask why-questions where they are unfamiliar with the background knowledge of scientific expertise. The same could be said of clinicians. Consider Kitcher’s and Salmon’s further discussion about the background knowledge (K) and its relationship to the questioner and respondent of why-questions: Thus, K may contain many propositions that neither the questioner nor the respondent knows. Moreover, S, [the questioner] may have false beliefs that are in conflict with propositions in K. S, [the respondent] may therefore offer corrective answers to flawed questions by pointing to items in K. 23 This raises an interesting point about background knowledge and generating explanations. When patients and clinicians ask for clinical explanations, a back-and-forth conversation between them is necessary to generate and clarify the components of the explanation. The position of questioner and respondent are not set, they are fluidic and navigated as the conversation advances. Also, the patient can be mistaken in what she asks for an explanation about; and the clinician can have grounds to edit the request, or to refuse to answer it. Recall the case I discussed earlier in this chapter of Rose, the maid, who was suffering from respiratory problems. She had been treated by a number of doctors before one discovered a more permanent treatment—a respirator to protect her from harmful fumes. It is during the conversations between doctors and patients when genuine shared background knowledge is generated. Rose was able to let her doctor understand the chemicals she is exposed to regularly. The doctor then is able to explain how these chemicals harm her lungs. In these conversations, clinicians and patients each ask the other to consider new information as possibly explanatory. A pragmatic explanation seems to be a preferred strategy to some acontextual explanatory approach, which can either ignore the patient’s true request, or dismiss the 22’ Kitcher and Salmon, “Van Fraassen on Explanation” 318. 163 request as irrelevant and just give the set CNI. This shared system sees the questioner and explainer as part of an epistemic community, which allows navigation/discussion/editing of K, rather than have it assumed as set. Also, a pragmatic approach to clinical explanations recognizes that medically-relevant knowledge is often wider than is scientifically-relevant knowledge.224 Applications in Clinical Practice Having laid out the apparatus of van Fraassen’s pragmatic explanations, I want to return to examples of clinical conversations about health. Can this explanatory apparatus improve doctor-patient discussions about health? If so, how? What would such conversations look like? Below are two hypothetical cases. The fictional conversations are italicized; my comments are interwoven and appear in standard text formatting. Racist boss and Uncle Bill: The use of Social Determinants of Health in R Recall the original case of Uncle Bill and Uncle Boy who described their health problems. Uncle Bill believed his workplace and his boss’s racist attitudes are strongly connected to his stroke, so Uncle Bill mentioned this to his doctor. The doctor provided a terse, indifferent response. This answer was both inadequate as an explanation and caused Uncle Bill to feel unheard as a patient with complaints. Below, I try to show how a conversation on this topic could be improved. Uncle Bill ( UB): Doctor, I never really understood what caused my stroke last year. Can you tell me, why did I have the stroke? 22" The point here is this: it may be that the background knowledge (K) in research explanations can be limited to that of “standard” domain of scientific knowledge. But in clinical explanations (between doctors and patients), the background knowledge used to generate an explanation will be much wider than the K for scientific explanations, and hence will not be easily limited. In fact, limiting K for clinical explanations without good reasons may lead to problems of miscommunication between patients and doctors. 164 ' VHE Here, Uncle Bill has initiated an explanatory request regarding his stroke. The CC he has chosen is { stroke, no stroke}. He has not explicitly stated his intended relevance relation. Doctor (D): Well, it’s hard to pinpoint any one thing, but it was likely a network of factors, including your blood pressure, your diet and exercise habits, you don ’t smoke but do drink. You also seem to have a history of this problem in your family. Didn ’t your brother have a stroke, too? Here, the doctor specifies the appropriate relevance relation to use here. A number of factors may be involved: there were possible genetic factors (given the history with the brother’s health) and the high blood pressure, and lack of exercise. So, it may be appropriate to consider a relevance relation of genetic causation, but Uncle Bill has something else in mind. UB: Yeah. But that bastard of a boss of mine...I really think he’s to blame. He hates black folks. He never had one nice thing to say to us. Even his own employees, he never cared if we lived or died. D: Well, I don ’t think I can say your boss caused your stroke, even if he ’s a racist. Not even if he called you every name in the book. He may be a horrible person, but he didn ’t cause this. Here, the doctor is trying to emphasize that there is no causal relationship between the names the boss called his employees and the etiology of the stroke. This choice of relevance relation shows no connection between the boss’s actions and Uncle Bill’s stroke. UB: Doc, that’s not what I’m saying. If words killed folks I have killed my share of people over the years. But my boss, he ’s still to blame. D: 0k, if not from what he said to you, then how? I ’m not sure I see what you ’re getting at. UB: That factory was a nasty place to work. The chemicals we worked with would burn our throats, our eyes. We complained, but nothing ever came of it. It made us feel sick to work there. But, we had to go to work We had families to feed, and that was the only job around. D: 0k, well, this I think I can talk about. Yes, exposure to those toxins at work may have contributed to your problems. But that was the chemicals, not your boss. 165 The CC is the same, but the RR may be the causal factors of chemical exposure that lead to the stroke. UB: Doc, you ’re still not hearing me. My boss knew those chemicals were making us sick. He chose not to fix the problem. He wanted to make his money, and he didn ’t want to spend a nickel to help us black folks out. If he hadn ’t been such a racist, he ’d have cared enough to clean out that factory, or put a working ventilation system in. As I see things, he made my brother and me sick. What is becoming clearer is that there is a miscommunication as to what RR should be used. The clinician has been using one of chemical and etiological causation. Uncle Bill seems to be using one about social contributions of disease. While there may be causal factors to these, the description cannot be only at the level of chemicals. It’s at a higher, social level, one that describes how racism negatively affects groups and individuals. D: 0k, I think I see what you ’re saying. I was thinking about the chemistry of your situation, what you were exposed to and how that may have caused your stroke. But you were thinking about related but more distant actions. It wasn ’t any one thing your boss said that made you sick, but a general pattern of how he treated you and others. Those were the actions that you think made you sick. I ’m not sure I can ever say with certainty that your stroke was caused by what you ’re suggesting. There is some scientific literature that shows a strong connection between certain disadvantaged groups—poor folks, some racial groups—and poor health. But that literature is looking at groups, and we ’re talking about something difi‘erent, about one individual of one of those groups. About the rejection of questions: In this case, we see that the doctor originally wants to reject Uncle Bill’s question about whether his boss made him sick. A pragmatic approach to medical explanations requires we first determine whether the question is a good question. The doctor originally acknowledges that Uncle Bill has made a valid request (asked a good question), but the doctor doesn’t believe he has data to support Uncle Bill’s hypothesis. So, the question was not rejected, but the doctor thought he had grounds to prove it false. Similarly, some questions may be acceptable (well formed), yet ultimately unanswerable due to insufficient data. 166 About the Clarification of questions: As Uncle Bill and the doctor talk, though, we see that their dialogue helps the clinician to edit and refine Uncle Bill’s questions. In this case, the intended relevance relation is made clearer. This allows for engaging the patient in developing the explanation, without the explanation becoming strongly relativistic. The clinician does not need to accept any and all suggestions from the patient; the clinician is able to edit questions to better examine topics that are within the domain of medicine, or topics for which we have data. About Social Determinants of Health as Explanatory: The doctor’s original RR focused on certain factors as more related to certain types of causal factors and patterns. Even if a complete explanation of the disease is possible (in the sense of Railton’s ideal explanations or Thagard’s CNI), it is clear that clinicians often focus on some of these features and set others aside?” So, Thagard may argue that “stress” could be a part of the CN I for a stroke, and therefore the CN I of a stroke already includes the factors Uncle Bill is worried about, i.e., stress on the job. But 1 have argued in Chapter 3 that seeing the features of CNI such as ‘stress’ or ‘environment’ as an all-inclusive category is problematic. Also, I have argued that Thagard’s CNI does not adequately account for social factors as contributing to disease. Insofar as clinicians do tend to utilize a CNI-like model of explanation, they are often blinded to these further possibilities. Even if such issues were taken to be part of a corrected CNI, there may be continued problems. Clinicians are trained to try to identify certain features of disease, e.g., viruses, bacteria, genetics, rather than others, e.g., poverty. Why this oversight, then? This is in part likely to the training of medical school and the limitations of clinical practice. Due to time and resource limitations, clinicians must make decisions to 22’ In the next chapter, I’ll discuss at greater length the values that might best influence such activities. 167 investigate some features and ignore others. Normally, this is a beneficial practice. But clinicians unreflectively narrow their fields of inquiry due to habit or due to use of an improper explanatory model. In the case above, the clinician was habituated to thinking about chemical and physical causal features; he had not considered (early in the conversation) the role of social forces upon Uncle Bill’s health. Thus, he was unable to hear correctly Uncle Bill’s questions. Consider that the doctor originally took the fact of the stroke suffered by the brother of Uncle Bill as proof of an inherited genetic predisposition. But, if we know that the brother also worked in the same factory with unsafe conditions, the brother’s stroke may have less to do with shared genetics and more to do with shared environments and social conditions. HIV and “Why am I sick?” Consider the case of Tom as an example where the clinician-patient conversation improves by clarification of the contrast class. Tom has been feeling ill over the past two weeks, so he seeks an explanation from his GP. The doctor runs a series of tests. She discovers Tom has TB, and she begins the proper course of treatment. Soon after this, as part of a follow up visit, the clinician must tell Tom that he is also HIV+. Tom: Hi, Doc. What have you figured out from the tests? Do you know what ’s causing this cough? Doctor: Tom, the persistent cough you have is due to Tuberculosis infection. We can treat that with antibiotics. Tom: Great. Doctor: But Tom, I also discovered another problem. Your blood tests came back positive for the Human Immunodeficiency Virus, or HIV. I’m sorry to have to tell you this... Tom: I don ’t understand. That can ’t be right. You just said that my cough was caused by tuberculosis! Here, there is a miscommunication about the contrast class. For Tom, the CC was {sick for the last two weeks, not sick before that}. HIV infection does not explain that 168 difference, since for it to show up in a blood test, infection must have occurred a number of months ago. What made this difference between these two time periods is the TB infection. This explanation, though, is of secondary importance for the clinician. As she sees it, the better explanation of the cough is Tom’s HIV infection. The TB infection is likely a result of Tom’s lowered immune status. In this situation, the time period used in the contrast class misleads Tom in how he thinks about his health. The more important explanatory event involved his becoming HIV+. The Case of J.S. Revisited Recall the case of J .S. from the Introduction: J .S. has felt ill for an extended period of time (at least ten years). Over the years, her symptoms have included muscular pain, fatigue, depression, and constipation. She has seen a number of clinicians and specialists, who have run a number of tests. Yet there is still not a clear diagnosis or consensus about why J.S. is ill. In fact, clinicians disagree about how to diagnose or label her condition. Reading through her chart over the years, notes point to various attempts: “chronic pain syndrome,” “irritable bowel,” and “somatization disorder?” In Chapter 4 I introduced a discussion of “medically unexplained disorders”, which are, in brief, “chronic and disabling conditions, presenting with extensive subjective symptoms, although objective findings or causal explanations are lacking.”226 A few common examples might be fibromyalgia, chronic fatigue syndrome, or temporomandibular joint disorder (TMJ). Medically unexplained disorders are interesting as a case for medical explanations, but they are also frustrating. First, identifying and treating these conditions is frustrating in that no clear causal network has been identified or described. This, by 22" Malterud, “Symptoms as a Source of Medical Knowledge: Understanding Medically Unexplained Disorders in Women” 604. 169 itself, does not make medically unexplained disorders a unique or surprising topic for medical explanations. It is often thought that for 20% to 40% of medical cases, no proper diagnosis or causal story is ever developed.227 Although these cases may be untreated (or forced into one diagnostic category or another), they often resolve. Because medically unexplained disorders are chronic conditions, their effects are long felt, and multiple conversations often occur between doctors and patients about their causes and possible treatment strategies. Second, these conditions are highly gendered: they affect women more frequently than men. Third, there is a strong correlation with sexual and physical abuse. Yet these similarities are not obviously helpful in generating an explanation of these disorders. Even if rape is connected to the history of many of these women, it is not clear how it is connected to their present symptoms. Most of these women have long since physically healed from the assaults, which may have occurred decades before symptoms arose. It is also not clear how gender is causally connected. Are these diseases genetic? Involved with social stresses that women face? J .S.’s primary care physician has become frustrated. He no longer knows what to suggest, what tests to order. He doesn’t know what to say when J .S. asks him what the cause of her discomfort is. As the case noted, he now feels exhausted, annoyed, and uncomfortable when talking with J .8. (although for the sake of professionalism, he tries not to let her know this). In brief, he is burnt out on this case. J .S. is also frustrated. The numerous appointments to see doctors and the tests they’ve done have not cleared up why she is sick. In fact, she remains just as confused now as she did before seeing the doctors. They’ve given her contradictory reasons and advice for how to treat the pain. She’s now suspicious of them, wondering if they really 227 Sarah Nettleton, Ian Watt, Lisa O’Malley and Philip Duffey, “Understanding the Narratives of People Who Live with Medically Unexplained Illness” Patient Education and Counseling, 56. 2 (2005): 205-2 l0. 170 In know what they are doing. She can sense they’re frustrated with her. But she feels trapped: if you can’t turn to your doctor when you’re sick, to whom else should you turn? Here, the doctor begins with something akin to Thagard’s CNI. What would a CNI explanation of J .S.’s case look like? It would involve a list (or diagram) of a number of different causal chains, which each may result in the cluster of symptoms J.S. is experiencing. A number of these have been eliminated by tests and other clinicians’ diagnoses. He’s eliminated viral and bacterial infections, food allergies, and other physical traumas. In this way, CNI is complementary of a differential diagnosis: we rule out certain possibilities. The remaining options warrant further examination to see if they can serve as an appropriate explanation. The doctor here feels that given all the tests that have been run, and the opinions of various specialists, that the possible physical causal aspects of J.S.’s case have been ruled out. So, what he is left with, and what he believes to be explanatory, is that J .S.’s condition is psychological. Her depression has mental, not physical, causes. Here, the differential diagnosis works much like the CNI. The doctor has eliminated various possible causes of her condition (or at least the ones he knows of and is able to test for). Believing he has ruled out most others, he believes the remaining thread—depression due to a mental disorder—is the appropriate explanation. We might question, though, whether the doctor has truly considered all the relevant information available to him. Certain causal threads are often disregarded by clinicians discussing medically unexplained disorders, but this pattern is common in that social determinants of health are often overlooked.228 The reasons for this likely vary given the circumstances of the disease and the clinicians’ training, yet some light is shed on this by a Canadian study of clinicians’ inquiries with patients with irritable bowel syndrome (one of the unexplained medical disorders). The study found that most clinicians thought routine inquiries into physical or sexual abuse should be made. Only 228 Ilnyckyj and Bernstein, “Sexual Abuse in Irritable Bowel Syndrome”, 801-805. 171 10% of doctors found such information to be “irrelevant to management.” While many clinicians favored the gathering of such information, only 23% always or frequently collected such data. Clinicians reported other factors for why they did not collect information on sexual and physical abuse, such as time constraints, and lack of resources for treatment referral (“If we can’t treat it, we should ignore it.”) Yet I agree with the authors of the study who write, “Patients have the right to understand causal or contributing factors to their disorders irrespective of treatment options.” As they rightly point out, numerous patients are informed of organ failure, even if their chances of receiving an organ transplant are minimal to nonexistent.229 The worry here about explaining diseases by reference to social causes also arises in the case of Uncle Bill. Recall there, clinicians may not cite racist social attitudes as “the cause” (or even as one of many causes) of Uncle Bill’s stroke because such things are often thought of as outside the domain of medicine. Yet such information, as I have argued, is often explanatorily important, even if medicine cannot correct or prevent such causes. It may be that in J .S.’s case, the doctor’s interests are shaping what avenues of testing he explores. He may not ask about her history with abuse and sexual assault, either because he does not believe it to be relevant, or because he does not know what he can do to treat it. So, the role of sexual assault may not be one of the causal threads in the doctor’s original CNI. This may be appropriate if we have strong reason to believe sexual assault is not causally related to something like irritable bowel syndrome. But the data, while not conclusive, seems to suggest that such assaults are in some way connected. The doctor may have chosen not to include this causal thread because he “can’t treat it.” But this, too, seems like a poor reason for excluding such information. 22” Ilnyckyj and Bernstein, “Sexual Abuse in Irritable Bowel Syndrome”, 801-805. The author’s quote continues “. . .in the case of abuse, patients may be encouraged by the link to their illness to help themselves or to seek help outside the health sector.” 172 At their next appointment, the doctor tries to talk out with J .S. the explanation he has developed. The conversation does not go well, though: J.S. rejects the explanation that her condition is the result of mental illness. J .S. says, “I’m sick, but I’m not mentally ill.” J .S. is thinking about her condition in rather different ways than is her doctor. She has had friends before who were depressed. Her situation, she believes, does not seem to be similar to their experiences. She wants to go to work, but feels too ill to do so. When her friends were depressed, they did not really want to get out of bed. J.S. notes that if she is depressed, it is a symptom, not the cause. She has lived in pain for so long maybe it makes sense for her to be depressed, but depression was not the start of her pain. J.S. has also been reading up about her condition on the web. She has found that a number of other women have had similar problems to her own. Many of them have been told they were “depressed”, it was “all in your head”, or otherwise made to feel they were crazy. But J.S. is clear that she doesn’t feel mentally ill. And here, J .8. may have at least some grounds to question this diagnosis. Women are diagnosed as depressed (and medicated) at a higher rates then are men, an understanding J .S. has developed from her own research on the web and conversations with other women. Many have wondered whether there is a true medical need here, or an unwarranted difference based in gender perceptions regarding how men and women are treated. Similarly, there is possibly a parallel history with Chronic Fatigue Syndrome. Before the bacterial causation was understood, people with chronic fatigue were often told their conditions were psychologically caused. By ruling out the known physical causes of that time period, doctors often thought this was the best possible explanation. Yet new data generated updated explanations. J .S. wonders if something parallel is happening here: scientists just need to discover what is going on in her body. What should the doctor make of J.S.‘s reaction? He has provided his evidence and his explanation. Is J .8. being obstinate in rejecting the explanation? Does she not understand some aspect of the explanation, or has the doctor indeed given a poorly 173 constructed explanation? One improvement to understanding this case is to understand that different types of explanatory work are appropriate here. Take J .S.’s understanding of depression, which she gleaned from the experiences of her friend. The doctor would be justified in explaining the friend’s depression, and why that caused her not to be able to get out of bed. The doctor can continue to explain that there are a variety of ways in which depression can manifest itself. What happened to the friend may or may not be similar to J .S.’s experiences. Notice here that this is to give an explanation of the friend’s depression (not of J .S.’s condition), yet this information is nonetheless explanatory to J .S. It may improve her understanding of her own case by clarifying misunderstandings about depression, and understanding the proper ways her case is analogous to her friend’s experience with depression. While miscommunications occur between doctors and patients generally, this case reflects common experiences of clinicians and patients involving medically unexplained disorders. The miscommunications lead both parties to fail to listen to each other, and to stop the search for an improved explanation of the case. As a means to improving understanding, I propose the following explanatory strategy. What corrections can be made to how J .S. and her doctor generate an explanation? Here, I propose that the activity of providing pragmatic explanations—in a transparent way—may help both parties understand where their miscommunication has occurred. This would involve both the doctor and the patient identifying the components of the pragmatic explanation that they are explicitly or implicitly using. Both the doctor and J .S. begin with the why-question, “Why does J .S. feel sick?” But because pragmatic explanations are contextualized, understanding the contexts from which they generate their explanatory responses will show why they are asking different why-questions. With the features of a pragmatic explanation identified, we can understand where miscommunication has occurred, where mistaken assumptions may have occurred, and what remedies may be possible. 174 What would the doctor’s pragmatic explanation look like? He might begin by identifying the why-question he and J .8. share. The question, “Why is J .S. sick?” can be broken down into the components of a pragmatic explanation: First, the contrast class: J .S. is {sick, not sick}. While this is a possible contrast class, as I argued earlier in the chapter, it is not the most informative. Having noted this in their conversation, they might revise it further: J .S. is {depressed, has bacterial infection, has viral infection, has food allergy, physical trauma}. Yet this contrast class has the danger of possibly blurring the distinction between CC and RR. There is a merging of the events to be explained (CC) and causal factors that bring about these events (RR). J .S. is {physiologically normal; suffering from depression, pain, fatigue (her current state); suffering from severe depression, pain, and fatigue such that she cannot function in her daily activities}. Here the doctor and patient can clarify the components of the explanation about which they are interested. The contrast may begin with the basic {sick, not sick}, yet this can be improved to be a more useful distinction about the degree of severity of J .S.’s symptoms. The process of articulating good why-questions will involve this sense of continuing refinement and reexamination. Note that an impediment in deve10ping the contrast class involves a difficulty in pinpointing the thing to be explained. That is to say, the contrast class for the explanation of this case is ambiguous. Because J.S.’s condition is not easily labeled or defined, there may be variation in how one articulates the topic. One possible way of resolving this is to make the contrast between the list of symptoms J .S. is experiencing, as opposed to other possible symptoms, or even a “health” state. A different explanatory strategy that would involve a different contrast class is to ask about the time period of J .S.’s condition. Why is J .S. sick {now, 15 years ago (the time before her symptoms began)}? The question then would focus on what made the difference between when she felt healthy and when she started to become ill. Although this is another possible explanation they could investigate, I will not develop it fully here. 175 var—fl : Relevance relations are filters that determine which causal factors will be considered as explanatory, and which will be set aside. We can ask what are the interests that influence the physician’s R in this case? The doctor here is not explicitly seeking to find only physical or physiological causes for J .S.’s condition. He is, though, strongly relying on the tests previously run and the expert advice of others, all of which are types of information allowing him to rule out various possible explanations as irrelevant to this - case. I have argued earlier in this chapter that generally clinicians are habituated into seeking certain types of explanatory evidence. First, they are more likely to look for m ‘3! -v_‘_flm - - . , recent (rather than distant) causes. Second, they are more likely to consider as causally relevant those incidents that they can treat or modify. While a wide range of factors may be considered, clinicians generally find it more helpful to consider information about which they can change or provide a treatment strategy for. We can ask patients to make their explanations explicit. J.S. begins with the question, “Why do I feel sick?” The topic she begins with is “I feel ill”, and she is using the basic CC {I feel ill, I do not feel ill}. Where J.S. differs greatly from her doctor, though, is in the use of her relevance relation. The doctor is using a relevance relation of eliminating possible causal factors. That is, he is trying to understand what is happening in J .S.’s body in the here and now. J .S., though, is looking for an explanation that stretches further back in time, asking about what was the original cause of her condition. She has also made it clear that she is rejecting the psychological explanation her doctor has put forth. I .8. may be making an important mistake here, just as I noted above that clinicians often make mistakes in explanations. J .8. begins with a (predetermined) relevance relation. She is denying a certain type of explanation, and perhaps she is doing so without good reason. It is one thing if her exclusion of depression is because she has a certain type of experience with depression and can eliminate it as a possibility. Perhaps if she herself had been depressed (or had been close to others with depression) she is 176 capable of making an informed, lay diagnosis. When asked why she believes this (or what is her evidence for this belief), she may respond, “Well, years ago, I had depression. This is not it.” If so, then she may have ruled out depression with a sort of folk- understanding (rather than with the diagnostic tests of a psychological expert). Although that scenario is possible, it does not appear to be what is actually occurring in J .S.’s case. Her own interests and background have prejudiced her against .. ‘ accepting an explanation of her condition involving mental illness. So, even though her I .l doctor’s evidence points to a mental disorder as explanatory, J .S. has rejected this _ possibility in advance. She is requesting a non-psychological explanation before she S knows whether such an explanation is possible. As I have argued, clinical explanations occur over time, they are edited in the back and forth conversations between clinicians and patients. Making the components of a pragmatic medical explanation explicitly stated, say as part of a medical chart, sets the stage to understand where disagreements occur. By making these components explicit, problematic explanatory strategies, dissimilar explanatory interests, or assumptions can be identified and addressed. For instance, J.S. may be premature in rejecting all explanations in terms of mental illness. But in writing down the components of her explanation, the doctor can point to this and discuss why he disagrees. He can also explain other things here, like the similarities and differences between J .S.’s depression and that of her friend. There will be multiple explanatory lines of thought possible in these conversations. If J .S. is concerned about the overmedication of women for depression, she can raise this worry with her doctor. He may, in fact, share this concern. If he does not, and rejects J .S.’s worry out of hand, then J .8. may choose to find a different physician to work with. Again, the first benefit in making the facets of a pragmatic explanation explicit is it allows both parties to put their position, their assumptions, and their explanatory interests “on the table.” This act of clarification allows for both self-reflection and critical engagement between 177 clinician and patient in generating clinical explanations. The patient wants to understand the doctor’s reasoning process, not just his diagnosis. She can thus make a more informed decision about whether to seek help from a different clinician if he is just unable (or unwilling) to hear her, or if he has dismissed certain evidence without cause. Even after the explanations are clarified, disagreements may still remain between J .S. and her doctor. Yet this need not prevent thoughtful consideration of treatment options. It may involve consideration of new information, or calls for further testing. For instance, the clinician may ask J .S. to speak with a psychologist to get an expert’s opinion about her psychological health. Even if the therapist says that J .S. is, in fact, depressed, the diagnosis may remain ambiguous. The therapist might say J .S. is depressed, but that she cannot tell at this point whether this is the cause of her pain, or whether the depression is an understandable result of living in pain for so long. Similarly, there may be a good reason for the doctor to reconsider the importance of J.S.’s history of abuse. It may be unlikely that J .S. will raise this information at this point if she has not already told him in the past; if the doctor has not asked at this point, it is unlikely he will ask now. But such information may come up in a discussion with a therapist. If so, the clinician may be more likely to consider it as part of his explanation. Making the components of a pragmatic explanation explicit may also benefit developing treatment goals for J.S. Recall the example van Fraassen uses regarding the multiple examples given by experts about why the car crashed. It may be that multiple explanations are possible about why J .S. is ill. A sense of uncertainty may remain in different ways the case can be explained. It may be, for example, that there is an underlying physical cause at play, one that has not yet been identified. But, it may also be that J .S. is suffering from depression, as her doctor currently believes. In the end, this disagreement will require further data of some kind—more tests of J .S., or perhaps better clinical research regarding medically unexplained disorders, generally. Yet J.S. needs treatment in the here and now. 178 One treatment strategy might be to keep open the various possible explanations, and then discuss what treatment options they imply. The doctor can suggest, “If we consider your condition to be a mental illness, then there are psychological remedies we should try. But if these do not work, say, after a period of 10 months, then we should reconsider your treatment options.” The decision to treat according to the psychological explanation (rather than the alternatives, a physical explanation, or a combination of both psychological and physical factors) here is not made by assuming one must be right and the other wrong. Instead, it acknowledges there may be many explanations possible for this case. The psychological explanation, though, allows for an easily identifiable therapeutic response: the standard treatments for depression. It may be more difficult to develop a therapeutic response based upon understanding J .S.’s case as having physical causes. This latter strategy may ultimately turn out to be correct, yet at this moment, it does not clearly identify the proper therapeutic strategy the doctor should prescribe. What, then, does an examination of the case of J .8. show about medical explanations? In this example, I have argued for the importance of making explicit the components of a pragmatic clinical explanation. Generally, physicians and patients involved with medically unexplained disorders report frustration. Both parties feel they are unheard by the other. By making explicit (say, by writing down as part of a medical chart) the components of the pragmatic explanations both parties are using, an improved conversation can occur. This strategy is a means to keep both parties talking and listening to each other. They can move towards mutual understanding by identifying the pragmatic interests that cause them to consider some information, but set other information aside. Gaps in the information available to them can be noted; these informational gaps may inspire further tests to be ordered. In the long term, further biomedical research may also fill in these gaps. Patients, too, have clearer ground for questioning or disagreeing with the explanations physicians provide. Finally, during a conversation about the pragmatics of clinical explanations, clinicians can show patients possible mistakes in their 179 explanatory requests. Such corrections, even when they involve explanations about persons other then the patient, are part of the explanatory conversations between patients and clinicians. That is, sometimes a good explanation will also include an explanation of what is not happening in the case at hand. Conclusion I have argued in this chapter that medical explanations are best understood as contextual, pragmatic explanations. In focusing on clinical explanations of disease (generated and shared by clinicians and patients), I have argued for using the structure of van Fraassen’s pragmatics of explanation. This strategy has many strengths. First, it requires that explanations be considered as contextual: what counts as a “good” explanation will be determined by the interests of the questioner, and different questioners can seek different explanations of the same event. Second, the resources of “contrast class” and “relevance relation” provide a foundation to improve explanatory requests. In practice, this can provide clarification of ambiguous why-questions. The act of clarifying the CC and R can also prompt clinicians to become self-critical about what information they are considering explanatory. This gives them a point to question whether this strategy is the best or the only explanation that can be provided. Similarly, articulating the contrast class and the relevance relation used within an explanatory response proves patients an entry point for joining the discussion; from here, they can help generate and evaluate the clinical explanation if they choose. Pragmatic medical explanations are therefore able to provide a theoretical context for both biomedical research explanations and for clinical explanations. Here, I have focused on clinical explanations to show that clinicians’ and patients’ different backgrounds (e. g., level of scientific expertise, lived experience of the disease) do not 180 prevent a collaborative clinical explanation process. In contrast to Thagard’s acontextual CN I explanations, pragmatic clinical explanations are not produced by clinicians and then provided to patients. Instead, clinicians act in conjunction with patients to generate clinical explanations. 181 CHAPTER 6: VALUES, INTERESTS, AND MEDICAL EXPLANATIONS Introduction In the previous chapter, I argued clinical explanations are best understood as pragmatic explanations, drawing from the work of van Fraassen. Such explanations are necessarily contextual. A good medical explanation responds to the explanatory needs of the inquisitor—that is, I argue for erotetic approaches to explanation rather than ontic. I have argued that patients and clinicians, as an epistemological community, are the epistemic agents of clinical explanations. Hence, good explanations are generated relative to the explanatory interests of the questioners, both patients and clinicians. This collaboration of patients and clinicians leads to the problematic question: whose interests should guide the explanation? This is a problem that van Fraassen had not addressed explicitly, largely because his examples of scientific explanations focused largely on peer-peer explanations (between fellow scientists). Clinical explanations, though, are not peer-peer. They are generated and used by patients (who are often scientific lay-persons) and clinicians (who have significant scientific backgrounds). It is generally accepted that values enter into scientific research.230 Yet there is considerable debate about which values ought to guide scientific research. Not surprisingly, then, this confusion carries over into which values ought to guide the generating and evaluating of clinical explanations. I argue for the necessary inclusion and examination of interests and values as part of medical explanations. Some authors (Kitcher, Thagard) have either (a) largely ignored 230 Richard Rudner, “The Scientist ‘Qua’ Scientist Makes Value Judgments”, Philosophy of Science 20 (1953): 1-6. 182 the role of values within explanations, or (b) have seen them as noteworthy only when considering the history of scientific and explanatory development. I begin this chapter by showing why values must be considered as part of (medical and other) explanations. I next consider whose values should be considered, since the epistemic agents involved in medical explanations will be different from those involved in biological or scientific explanations. Clinical explanations ought to be guided by scientific research, but not to the exclusion of patients’ interests and values. What is needed is a set of values that can structure clinical explanations such that they are contextual (erotetic explanations), will retain scientific rigor, and will meet patients’ (often non-scientifically guided) interests. Kuhn argued for a number of scientific values (i.e., accuracy, consistency, scope, simplicity, and fruitfulness), and I begin with these as a possible basis for values that ought to guide medical explanations. I find these of limited utility, though, for clinical explanations. They too quickly lead to explanations that ignore patients’ needs. I argue for a solution by applying the work of Helen Longino. Longino has argued for a set of explicitly feminist scientific values, i.e., empirical adequacy, novelty, ontological heterogeneity, complexity of relationship, applicability to current human needs, and diffusion of power. Like Kuhn’s list, Longino’s listed values are meant to be “epistemic values”—that is, that reliance on them tends to improve the chances that the judgments based on them are (at least approximately) true. I favor Longino’s list, though, in that it explicitly considers social dimensions of science and medicine. Such values guide medical explanations as drawing from scientific research, yet they differ from Kuhn’s values in important ways. Longino’s values intentionally address the political and contextual needs that shape scientific research, and they begin with the assumption that scientific research is socially situated. I then show how such values help “keep explanatory space open” in clinical explanations. I apply these feminist values to a number of cases, including J .S. and the category of unexplained medical disorders, and 183 show how Longino’s set of values is more successful for clinical explanations than those of Kuhn. Longino’s set of values, when incorporated into clinical explanations, both maintain scientific rigor and make clear the normative/political aspects of medical explanations. These social/political aspects are wrongfully veiled and ignored in Kitcher’s and Thagard’s work, yet are indispensable to proper medical explanations. This sets the stage for claims that generating medical explanations is both an epistemic and a normative/political activity. The Role of Interests in Explanations Proponents of causal explanations often argue that “correct” or “complete” explanations are those with an extensive list of casual factors or a detailed causal web.23 ' By describing all the causal factors of a given phenomena, we have ‘explained’ it. Salmon does this, requiring that we understand the mechanism of causation. Peter Railton’s ‘ideal explanatory text’ is another version.232 Thagard takes a similar approach with explanations as causal network instantiations (CND, arguing that the levels of causal detail provided in medical explanations are relative to the situations. This causal listing is seen to be a complete story of how an event came about, and thus is explanatory. The act of explaining is independent of value judgments in that generating an explanation is seen to be unaffected by individual’s interests. Thagard implies that values exist in biomedical research, but in the domain of the development of scientific programs and biomedical research. For instance, homophobia may have been a social factor that limited (or otherwise adversely shaped) early HIV research. Yet, the CNI does not describe any values of the inquisitor 23' Thagard’s CNI is one example of explanation as a causal web. 232 For more on Railton, see Wesley Salmon, Four Decades of Scientific Explanation, 159- l 63. 184 as being present. So, for Thagard a medical explanation does not recognize or describe these values, nor show how they structure the explanation. In contrast to Thagard’s avoidance of values within explanations, van Fraassen argues that a list of (only) causal factors, even if complete, can never be considered explanatory. For van Fraassen’s pragmatics of explanations, interests and context of the questioner shape explanations; a complete list of causal information lacks critical ,.‘ information that is relevant to the production and evaluation of explanations. Making just this point, van Fraassen argues that causal facts are necessary but not sufficient components of explanations: The description of some account has an explanation with respect to a certain relevance relation and a certain contrast-class. These are contextual factors, in that they are determined neither by the totality of accepted scientific theories, nor by the event or fact for which an explanation is requested. It is sometimes said that an Omniscient Being would have a complete explanation, whereas these contextual factors only bespeak our limitations due to which we can only grasp one part or aspect of the complete explanation at any given time.233 At this point, I take it that a medical researcher’s CNI explanation is an inferior version of the Omniscient Being’s ‘complete explanation’. (The Omniscient Being would be able to provide ideal explanatory texts for all phenomena, yet human clinicians often can provide much less.) Although Thagard may admit that most medical explanations are not yet fully complete, their ultimate goal would be to achieve such a complete explanation. For CNI, having a more-detailed causal web makes an explanation superior to a less- detailed web. That is to say, explanatory ability increases in proportion to our understanding of the causal complexity. Explanatory value is in the ability to discover and describe the complex set of causal factors. But van Fraassen rejects this goal: 2” Van Fraassen, The Scientific Image, 130. 185 But this [goal of a complete causal list] is a mistake. If the Omniscient Being has no specific interests (legal, medical, economic; or just an interest in optics or thermodynamics rather than chemistry) and does not abstract (so that he never thinks of Caesar’s death qua multiple stabbing, or qua assassination), then no why-questions ever arise for him in any way at all--and he does not have any explanation in the sense that we have explanations. If he does have interests, and does abstract from individual peculiarities in his thinking about the world, then his why-questions are as essentially context-dependent as ours. In either case, his advantage is that he always has all the information needed to answer any specific explanation request. But that information is, in and by itself, not an explanation: just as a 3pcrson cannot be said to be older, or a neighbour, except in relation to others.2 Here, we are reminded that the Omniscient Being is dramatically different from us as humans. Hence when the Omniscient Being is used as the (ideal) subject who generates explanations, such an agent is a poor substitute for theorizing about the explanatory needs of (real) persons. The Omniscient Being has no interests; as such, it can neither produce nor judge explanations, since explanations (by definition) are more than just factual information. Explanations can be produced only in conjunction with specific interests and values, values that are inaccessible to the Omniscient Being. Human beings, though, do have the necessary interests and values to produce explanations. These same values will change between contexts. As such, medical explanations and scientific/biological explanations may involve different interests and values. In the sections below, I try to sort out the values that should be considered in the formation of medical explanations. If van Fraassen is right, then an interest-free strategy to medical explanations cannot work. The interests and values of the inquisitor necessarily shape an explanation. The goal of an explanation is to get clearer as to what counts as good reasons for considering some facts (rather than others) as explanatorily relevant. In this way, the interests of the inquisitor determine the relevance relation, a key feature of pragmatic explanations. Within the Context of discussing clinical explanations, this leads us to ask: ”4 Van Fraassen, The Scientific Image, 130. 186 whose and which interests should shape clinical explanations? What should occur if clinicians and patients have different explanatory interests, which often will be the case? This will be the topic of the next section. How are Interests and Values Important to Medical Explanations? For a successful pragmatic theory of medical explanations, the interests and values of the inquisitor must be identified. To identify which values are appropriate in medicine, I will first explore the role of values in scientific research. If values play an important role in the construction of scientific explanations, then which values (or which types) need to be considered? The literature surrounding values and scientific inquiry often questions whether values and interests play a role in scientific projects. While some have argued that scientific programs are value—free, many authors have argued that science, as a social research program, cannot escape from certain values?” Disagreement remains, though, as to which values are appropriate. The general worry about allowing values (especially personal or contextual values) into explanations is that they quickly may lead to relativism. If some set of constitutive values, though, can be identified as part of scientific projects, explanations can influenced by values, yet avoid relativism. Authors interested in identifying such interests or values in science often distinguish between epistemic and non-epistemic values.236 Epistemic values are those that are fundamental to the doing of science. Reliance on such values tends to improve the chances that the judgments based on them are (at least approximately) true. For instance, ‘consistency’ is one value thought to be important to scientific research; 2” See, for example, my comments on the works of Richard Rudner and Mary Ann Cutter. 23‘ Throughout this paper, I will use the terms epistemic values and constitutive values as interchangeable, and non-epistemic, personal, or political values as roughly interchangeable. Only when I make note of this should any philosophical difference be considered in my wording. 187 scientific theories that are consistent with the predictions of other related, accepted theories are thought to be better than those that directly contradict currently held beliefs. Non-epistemic values are not fundamental to the doing of science, and they change given the interests and context of the inquirer. Debates about which scientific research programs to fund, say, the Human Genome Project versus manned exploration of Mars, are value judgments made about scientific research. But the judgment is not one internal to the doing of science, but about what scientific projects ought to be funded. Described another way, values are often divided between those essential to the doing of science versus those that may exist within the scientist but are not an essential component of the scientific program. (Specific examples of these values will be provided below.) Assuming for now this is a valuable distinction, we can ask what category of values van Fraassen is discussing when he describes the interests and values of the inquirer as shaping explanations. I would argue that both epistemic and non-epistemic values influence the construction of relevance relations (which types of causal factors are cited) and contrast classes (which possible events are to be considered). The non- epistemic values may be personal to the inquirer, and this part of any explanation may change given the person seeking the explanation. But, it may be that certain constitutive, non-contextual, epistemic values must be considered as well as part of scientific and medical explanations.237 For van Fraassen, the only distinction between scientific and non-scientific explanations is that scientific explanations draw upon science: To call an explanation scientific, is to say nothing about its form or the sort of information adduced, but only that the explanation draws on science to get this information (at least to some extent) and, more importantly, that the criteria of ‘37 For example, a patient, when asked about his use of herbal remedies, may explain that St. John’s Wort relieves his depression. It is an inexpensive remedy that he believes helps him control a part of his life. His doctor, though, may be slow to prescribe this activity, since there is little evidence supporting the efficacy of St. John’s Wort in treating depression. Thus, the doctor is relying on scientific values of accuracy and consistency, and these are in conflict with the patients’ valuing of personal experiences and his sense of control over his own body. 188 evaluation of how good an explanation it is, are being applied using a scientific theory.238 Distinguishing between scientific and non-scientific explanations is not always an easy task. Such distinctions are made more difficult for my project of understanding what values (constitutive and contextual) influence the structure of medical explanations. Below, I describe in greater detail the personal and then constitutive values that shape explanations, both scientific and medical. Following that, I explore possible constitutive values for medical explanations, and show their influence on shaping those explanations. Such constitutive values ought to be shared by both clinicians and patients in generating clinical explanations. Personal Values Investigating and identifying personal values has an important role in critiques of science. Scientists have their own personal interests in their research, such as excitement in a given topic, or desires for wealth, fame, or tenure. The scientific community at large expresses certain values in its practice of scientific review and peer-criticism regarding scientific publications. Other interests come into play when we look at those funding scientific research, e. g., the economic interests of drug companies, or the interests of individual citizens when the government supports research projects using money collected through taxes. In large part, concerns about tenure or profit are considered to be non-epistemic values. Although these may be real interests of researchers, such interests are not essential to the doing of scientific research. In the formation of scientific explanations (a peer-to-peer project), these personal values may be shared or rather different between two scientists. Clinical explanations differ, though, in that they are non-peer explanations. Hence, the contextual components of clinical explanations are immediately more ”3 Van Fraassen, The Scientific Image, 155-156. 189 complicated than that of scientific explanations. This complication is due to the number of different types of persons involved: clinicians, medical researchers, patients, and possibly legal experts and bioethics experts (ethics committees) involved in health care. Scientific explanations and most research medical explanations are generated as peer— peer interactions; scientists generate explanations typically to share with other scientists. In contrast, clinical explanations typically are shared between non-peers. Clinicians and patients both utilize them, as I argued in Chapter 3. So, the overall set of contextual values is much larger in medical explanations than in scientific explanations. Medical researchers may still have the same personal interests as scientific researchers (tenure, fame, money, pride, enthusiasm for work, etc.). As clinicians, doctors and nurses will be under additional constraints, especially regarding the allocation of money and time that can be afforded to each patient, and the general desire to care for patients. Patients also will bring a wide range of personal interests to clinical explanations. Often, these may conflict with the epistemic values of medical research. In other words, there may be times when patients’ personal or political interests in seeking explanations may conflict with the scientific values involved. As such, if we can show that there are constitutive values to medical explanations, we may have grounds for showing that some patients’ interests (and some clinicians’ interests) cannot be taken into account as part of a medical explanation. Constitutive Values Thagard’s explanatory theory is incomplete in that it cannot identify values that are at work in medical explanations. Thus, this theory cannot provide grounds for critiquing these values as either proper or improper. Without some common set of epistemic values, then it will be difficult to navigate differences between patients’ and clinicians’ goals. Why does this matter? I have tried to argue that a model like Thagard’s 190 CNI may not meet the needs (that is, may not reduce the puzzlement) of patients seeking medical explanations. Without common values (epistemic values) between clinicians and patients, then the patient is probably “just out of luck.” The medical explanation is the best that the clinician can provide; if that is not enough, then the patient will have to seek a different source to reduce her/his puzzlement. But, if there are common values, then we can begin to seek medical explanations that help both parties. It is reasonable to believe that we could come up with a list of the values that are utilized in explaining medical phenomena. To begin this list, I pull from literature on values within scientific programs in general, since providing (scientific) explanations is one of the common goals for science. I choose this strategy in large part because if we can identify constitutive values of medical explanations, this gives clinicians and non- clinicians shared grounds for explanatory conversations. Is the list of constitutive values within medical explanations different from that within scientific explanations? One might argue that medical explanations need to be understood as scientific products, free of social-political influences. If so, explanatory values could be generated from the “scientific” aspects of medicine; explanations would thus be independent of things that are often described as the social (or artistic) aspects of medicine: clinical/professional attitudes, how to interact with patients, etc. But this strategy again assumes a strong line separating epistemic and non-epistemic values, a line of thought I am not willing to support here. Some have argued that this distinction between constitutive values and personal values in science is not useful since the line between epistemically valuable and politically valuable considerations is often unclear.239 To hold to this distinction strongly assumes that there is some set of values that is fundamental to scientific inquiry, and separable from researchers’ personal interests. Many have argued that such an example is 239 Phyllis Rooney “On Values in Science: Is the Epistemic/Non-Epistemic Distinction Useful?” PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association 1 (1992): l3-22. 191 “accuracy”, which ought to be considered an indispensable, epistemic value of science, which will be discussed below. Any scientific research project that cannot match its theories to its observations cannot be considered a successful scientific project. Yet others, including Helen Longino. argue that some political interests can actually become constitutive values in science, and that standard scientific values have goals (perhaps veiled) about how scientists and scientific research will influence the world. h Below I look at a longer list of values that may be considered essential to scientific research. But the question still remains about whether it is useful (or possible) to distinguish between epistemic and political-personal-contextual values. After 'WT ' describing Kuhn’s list of constitutive scientific values, I look at Helen Longino’s list of feminist scientific values. I believe this does two things for this conversation: first, it shows that the line between constitutive and political values in science is not as sharp as some had hoped. Second, it also provides a set of constitutive values that both clinicians (as scientists) and patients (as non-scientists) can share when generating medical explanations. If these parties can share these values, then there is less demand for merging patients’ and clinicians’ non-epistemic values. Kuhn’s ‘Standard’ Scientific Values In his often cited 1977 article, “Objectivity, Value Judgments and Theory Choice”, Thomas Kuhn discusses the values that proponents of rival scientific paradigms shared (despite the significant differences that separated them during scientific revolutions). These values are accuracy, (internal and external) consistency, scope, simplicity, and fruitfulness. Kuhn argues these are the values scientists use to guide their judgments and when making choices between rival theories and explanations. At a minimum, this is a common set of values used by scientists, even those coming from highly diverse backgrounds. For instance, there is a general scientific consensus that 192 quantum mechanics is in some way better than Newtonian physics. This is largely because the former allows scientists to make predictions that are more accurate and more consistent than the latter theory. Kuhn argues that these five values are constitutive of all scientific projects. As such, these values are integral to scientific projects, and focusing on them avoids the irrational (non-scientific) aspects of personal/political/contextual interests. “To the extent that arguments can be based on these shared values, scientific revolutions are rational.”240 Yet Kuhn denies that these values can be rank-ordered according to their importance. These values require further clarification, and they may not all be able to maximally implemented at the same time. Also, there is no means for choosing between two theories that equally implement these values. When making choices between rival theories, “cognitive values are ultimately a matter of subjective preference that transcends rationality, and that non-rational psychological and social factors must play a vital role in determining which theory wins the allegiance of the scientific community.”241 These values work to organize scientists to keep them rational, despite other personal and methodological differences. These values shape scientific practice, including explanation formation. Other authors have proposed other value sets that are necessary for scientific explanations. Eman McMullin “reworks” Kuhn’s list: predictive accuracy, internal coherence, external consistency, unifying power, fertility, and simplicity.242 Helen Longino lists the standard constitutive values, those “governing values and constraints. . .that. . .are generated from an understanding of what counts as a good explanation” are as follows: truth, accuracy, simplicity, predictability, and breadth.243 24° Martin Curd and J. A. Cover (eds.), Philosophy of Science (New York: Norton and Company, 1998), 84. 2" Curd and Cover, 84. 2’2 Eman McMullin, “Values in Science”, PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association, Vol. 1982, Volume Two: Symposia and Invited Papers. (1982), pp. 3-28. Phyllis Rooney discusses McMullin’s list in “On Values in Science: Is the Epistemic/Non-Epistemic Distinction Useful?”, 14. 243 Helen Longino, 1990, Science AS Social Knowledge, 4. 193 Now, Kuhn’s, McMullin’s, and Longino’s concepts of scientific values are not identical, and for my argument here they need not be.244 What is interesting is that these authors have rather similar lists, yet none see their list as the single correct set of values possible?” For the rest of this section, I briefly describe each of Kuhn’s constitutive scientific values. Although other authors have provided various alternate lists of standard scientific values, for now, I am going to discuss Kuhn’s since these have been widely discussed in the literature, and they are (for the needs of this discussion) similar enough to the proposed alternatives. 1) “Accuracy”: This is a requirement that observations, experimental data, and experiences match up with a theory. This is related to what other authors classify as empirical adequacy. In the case of rival explanations, the preferred explanation will have the maximal correspondence between what a theory predicts and the data collected. 2) “Consistency”: Consistency has two distinct facets: internal and external. Internal consistency of a theory is a function of its internal logic; the theory should avoid internal logical contradictions. External consistency is the degree to which this theory avoids contradictions with other currently accepted theories in related scientific fields. 3) “Scope”: A theory’s scope refers to the range of phenomena that the theory can account for or can explain. Kuhn argued that an explanation should explain more than the phenomena at hand by supporting or confirming other theories and more distantly related phenomena than just the originally investigated explanandum. Explanatory success should be both immediate and further reaching. Note that in my discussion in Chapter 2 of the range of phenomena and patterns that can be considered explanatory. Some philosophers have argued good explanations ought to cover a wide range of phenomena, such as Philip Kitcher’s use of unification. 2‘“ In fact, they may lead to different conclusions, different outcomes as part of scientific research. 2’5 See Rooney “On Values in Science” for arguments to support that claim. 194 ”it 4) “Simplicity”: This requirement or value for scientific explanations can be interpreted in a number of different ways: simplicity of language, simplicity of ideas such that they can be understood, minimal number of assumptions, minimal premises of a given theory.246 When presented with two explanations of the same event, simplicity favors the explanation with fewer complex mechanisms or assumptions. As Longino writes, “The simpler theory is one that stipulates fewer [causal] entities or fewer processes?“7 Note that the requirement of “simplicity” is not the same as “easy to understand.” In many medical contexts, physicians recommend giving patients, the least- complicated, least-jargonistic explanation that can be provided in order to improve communication. 5) “F ruitfulness”: As an epistemic value, fruitfulness is the promise or the appearance of filture productiveness of a theory. It is not a value concerned with empirical aspects of the theory. Fruitfulness is about counting the number of new discoveries resulting from a given explanation; but instead about the appearance or likelihood that (as it appears now) this explanation will lead to future discoveries and connections. Longino’s Values of Feminist Science Feminist critics of science have generated alternate values for feminist science. Among these authors, Helen Longino argues for the following values to a feminist science: empirical adequacy, novelty, ontological heterogeneity, complexity of 248 relationship (at times called mutuality of interaction) , applicability to current human 7'“ Rooney describes in great detail the possible readings of simplicity. 2’7 Longino, “Cognitive and Non-Cognitive Values in Science: Rethinking the Dichotomy”, in Nelson and Nelson, Feminism and Philosophy of Science, 43. 243 Longino, “Cognitive and Non-Cognitive Values in Science”, 47. 195 needs, and diffusion of power.249 These feminist values need not be considered the one and only set, just as Kuhn admits his set of values may be subject to revision and further interpretation. Also, not all of these are exclusively “truth orient ” values, but Kuhn’s simplicity is not truth oriented, either. These feminist scientific values, as a set, have both truth and pragmatic/political goals. Longino’s point is “not to endorse these [feminist virtues] as substitutes for the conventional ones, but to use their advocacy by others to show how seemingly disinterested epistemic criteria can serve political ends.”250 They seek to bring about scientific knowledge that does not generate oppressive research strategies. 1) Empirical adequacy. Empirical adequacy may be a requirement for any reasonable explanatory schema. As such, Kuhn and Longino similarly require as a facet to scientific projects some strong connection between theory and observation. Kuhn’s accuracy and Longino’s empirical adequacy are both important scientific values in that science ought to bring us closer to understanding. Any attempt at a medical explanation must strive to be at least approximately correct: Empirical adequacy is valued for, among other things, its power when guiding inquiry to reveal both gender in the phenomena [being studied] and gender bias in the accounting of them. . ..Failure to meet the standard in a strong sense, i.e., the generation of statements about what will or has been observed that are incompatible with what has actually been observed, is grounds for rejection of the hypothesis or theory in question.25 ' Why is empirical adequacy a feminist value, rather than a general constitutive value of all science (as Kuhn had argued)? Many feminists have used this to discredit scientific theories that “purport to show a biological etiology for differences ascribed on 2’9 Although Longino has used these values in a number of articles, she first argues for explicitly feminist scientific values in “In Search of Feminist Epistemology,” Monist, 77 (1994): 472-486. 25° Helen Longino, “Interpretation Versus Explanation in the Critique of Science”, Science in Context 10. l (I997): ll7. 25' Longino, “Cognitive and Non-Cognitive Values in Science”, p 45. 196 m the basis of sex.”252 For instance, scientific projects valuing empirical adequacy can work to dispel sexist myths, such as women are biologically determined to be ‘the weaker sex’, or that women cannot be as intelligent as men, or that women’s emotional states are due to the location of wandering uteruses (hysteria). One problem in using this value is determining how much accuracy constitutes ‘enough’ accuracy. Such judgments are made by determining the usefulness of the (approximately correct) knowledge as judged against the dangers that may result from these errors. This judgment, though, is a pragmatic decision made by those involved, and such decisions will vary given the circumstances and those involved. Does the move away from “accuracy” to “empirical adequacy” say something about the judgment made here? Mainly, it says that we will accept a bit more inaccuracy, if we can get a good enough theory, and it meets the other feminist scientific values (described below), such as applicability to human needs and novelty. A different tack for understanding empirical adequacy (rather than accuracy) as a feminist value is that feminist science may worry less about eliminating all error, and instead see other values as equally important; Kuhnian values may over-emphasize the importance of accuracy. Similarly, empirical adequacy may be more helpful than “accuracy” to clinical contexts because often factors in clinical practice (e. g., time limits) may limit the amount of data that can be collected. In such circumstances, accuracy may affect the constraints and consequences patients are willing to accept. For example, in some cases, patients may be willing to accept a greater level of inaccuracy, as long as the harms are not excessive. But in examples with greater severity of consequences, patients may expect a greater level of accuracy (or be less forgiving of inaccuracy). We can also admit that some harm is inevitable as errors creep into our explanations. Since absolute truth may be impossible to 2” Longino, “Cognitive and Non-Cognitive Values in Science”. p 45. 197 gain, we will necessarily face situations where we will have to accept mistakes. A value judgment about what risk of error we are willing to accept will have to be made. 2) Ontological Heterogeneity. Ontological homogeneity, or uniformity, “characterizes theories that posit only one sort of causally efficacious entity, or that treat apparently different entities as versions of a standard or paradigmatic member of the domain, or that treat differences as eliminable through decomposition of entities into a single basic kind?”3 Longino argues for ontological heterogeneity as a feminist remedy. This scientific value favors expanding the ontology used within scientific research. This value favors expanding the types of causal factors beyond what typically has been considered, and critically examining our reliance upon only some features. For example, E.F. Keller’s work on DNA as ‘master molecule’ has tried to show that some conceptions of genetics overlooked the complex different causal factors involved in trait formation, prioritizing the work of genes. This also leads to an overly simplistic understanding of the interactions involved. 3) Novelty. “By novelty, I understand models or theories that differ in significant ways from presently accepted theories, either by postulating different entities and processes, adopting different principles of explanation, incorporating alternative metaphors, or by attempting to describe and explain phenomena that have not previously been the subject of scientific investigation.”254 Related to ontological heterogeneity, we see here that novelty is the move to incorporate theories and phenomena previously overlooked by standard scientific programs; often these overlooked issues are those associated with the lives of women and other minorities. Novelty could, of course, have stronger and weaker interpretations. The strong interpretation demands new frameworks and theories to replace current ones in the domains in which they are currently employed. On the weaker interpretation, 2” Longino, “Cognitive and Non-Cognitive Values in Science”, p 46. 25" Longino, “Cognitive and Non-Cognitive Values in Science”, p 45. 198 new frameworks are sought in satisfying a demand for scientific understanding of hitherto neglected phenomena.”5 What is implicit, though, in both the stronger and weaker forms of novelty is that scientists must first critically examine the scientific and social processes that brought them to this current research. In order to seek novel explanations and theories, they must seek how past explanations / events erased theories and topics important to women’s experiences. Only then can a strategy to avoid this pattern be developed--a novel research strategy. An example of this is provided in later sections, where I describe how a pragmatic model of explanation works to achieve the value of ‘novelty’ by providing a different way of understanding medical explanations. Embracing novelty does not diminish the requirement of empirical adequacy. New theories may/will provide better accounts of different phenomena, and shifting the focus of a new theory may cause existing data to become irrelevant or to disappear. 4) Complexity of Relationships / Mutuality of Interaction. Part of the goal here is to facilitate the formation of (often causal) models that identify systems that have historically valued/enabled male-domination in scientific research (such that these might be later avoided). Relationships are valued when seen to be interactive, rather than unidirectional, multi-factored rather than single-factored. Reproductive explanations, those emphasizing complexity of relationships, have been developed; these replace “models of energetic sperm acting on passive eggs with models of mutual interaction [revealing] the egg’s considerable contribution to the process of gametic fusion.”256 Other explanations that emphasize this value are those genetic explanations that try to show the extremely complex relationship between genes, cellular environment, and other types of environments on the trait formation. These models thus reject ‘the master molecule’ conception of genetics and trait production. 2” Longino, “Cognitive and Non-Cognitive Values in Science”, p 46. 2” Longino, “Cognitive and Non-Cognitive Values in Science”, p 48. 199 5) Applicability to Current Human Needs. This criterion values the avoiding of scientific projects that seek static knowledge, that is, knowledge for its own sake. Also, domination as a trend should be avoided (or less-highly valued), so military research for the goal of domination is to be avoided. Instead, scientists first should seek programs that meet the needs of current persons. This may mean undertaking scientific projects that seek elimination of hunger and disease, rather than the exploration of Mars. Medicine is fundamentally about addressing current human needs. Part of what separates medicine from other sciences, it may be argued, is that medicine is an applied science that focuses on improving health of real people. Patients, in face-to-face meetings with clinicians, require explanations about their symptoms or their prospects for treatment. Even in the research components of medicine, the research is undertaken in order to improve patients’ health and to prevent future illness. Rarely is medical research done for-knowledge-sake-alone.257 So, in this case, the value of “applicability to current human needs” is already a value in medicine. It is likely that decisions about funding competing research programs may favor, say, curing HIV over curing smallpox. But that is about choice of research programs, and in understanding medical explanations, I am here interested in theory choice. Later, I describe the role this value has in the generation or evaluation of clinician-patient (face-to-face) explanations. 6) Diffusion of power. Longino describes this value as the ‘practical version’ of ‘mutuality of interaction’. It favors models that utilize mutuality rather than dominate- subordinate roles. This value “gives preference to research programs that do not require arcane expertise, expensive equipment, or that otherwise limit access to utilization and participation.”258 This virtue does not try to maximize accuracy, but instead maximizes 257 Munson, “Why Medicine Cannot be a Science.” 25" Longino, “Cognitive and Non-Cognitive Values in Science”, 48. 200 h. the number of people who have access to research programs (by minimizing features that prevent such access). Diffusion of health care can be implemented within health care contexts. “Feminist health professionals urge a preference for medical practices and procedures that empower the individual woman either to make decisions about her health or to retain control over her own body.”259 Debates regarding the medicalization of childbirth in large part reflect this value: critics argue for allowing women to control of degree of technology involved in childbirth, which goes against national trends favoring (seemingly mandating) high-tech interventions, even for normal births. Diffusion of power is possible both at the practical and the theoretical level. At the practical level, as Longino notes, we could focus on “hygiene rather than high-tech interventive measures available only to the few.”260 At an abstract level, though, diffusion of power may occur when allowing both clinicians and patients to share the power to generate and evaluate medical explanations. This would involve seeing both clinicians and patients as epistemic agents worthy of listening to, but they need not be considered equal in all conversations. Patients must be heard, shown why incorrect, and not immediately dismissed.261 Why are these six values described as ‘feminist’ values, or values of feminist science? Longino rejects claims that these are values resulting from a feminine or female orientation to the world. These virtues of a feminist science “are neither uniquely nor intrinsically feminist, but. . .feminists could argue that theories exemplifying them would be more likely to satisfy feminist cognitive aims (which are also socio-political aims) — 2” Longino, “Cognitive and Non-Cognitive Values in Science”, 49. 26° Helen Longino, “Gender, Politics, and the Theoretical Virtues”, Synthese 104 (1995): 389. 2‘" This theme of diffusion of power—about the role of patients in medical explanations—is obviously tied to my comments in Chapter 4 about patients’ roles as part of the epistemic communities that generate clinical explanations. Part of my motivation in that chapter was to elucidate a means by which patients were considered important and primary agents contributing to the generation of clinical explanations. Although they are part of the epistemological community that generates medical explanations, this does not mean they have identical roles (or identical power positions) as do clinicians. Yet the move to include patients as epistemological agents is to empower them in a way that Thagard’s CNI does not. 201 namely to make women and female-identified phenomena as well as gender relations more visible.”262 She also denies that the reason these values are “feminist” is because women are often outsiders or underrepresented within the sciences. Instead, Longino argues these values are feminist scientific values for more pragmatic reasons, mainly what these values can do for feminist inquiry and critique of the sciences. “Revealing gender in a feminist context means revealing an asymmetric power relation that both conceals and suppresses the independent activity of those gendered female.”263 The project of revealing gender within the sciences and medicine is not an attempt to identify gender and gender biases as ubiquitous. Instead, Longino’s project here is to show that the standard values of scientific inquiry typically work to mask or to hide gendered discrepancies of power, while the proposed list of feminist virtues are more likely to identify and less likely to promote such gendered power discrepancies. As such, feminist values are more likely to point out (and not reinstantiate) social injustices through scientific programs. Medicine, as responsible for minimizing social harms and improving patients’ health, will benefit from taking seriously these values within medical practices. Why does feminism want to engage with science and medicine as a means for liberation from oppression? For one thing, science and medicine are both potentially powerful allies in the struggle to improve women’s rights. Science can expose and denounce poorly thought-out positions, and discredit their attempt at scientific credibility. For example, science can be a tool to dismiss oppressive views about the supposedly “innate” differences between the sexes: that men are more rational, women more emotional. Scientists, working from a different perspective, may also hold a type of status, credibility, and moral authority such that their claims are more convincing to the general public. 262 Longino, “Cognitive and Non-Cognitive Values in Science”, 5 l. ‘63 Longino, “Cognitive and Non-Cognitive Values in Science”, 50. 202 My project of theorizing pragmatic medical explanations should not be properly called a feminist project (in any strong sense) since I am not beginning with the explicit goal of revealing women’s oppression within science. I have not begun my inquiry into the nature of medical explanations with the explicit claim of seeking improvement to women’s lives. Still, I find the feminist critiques of science and medicine strongly informative and worthy of consideration. These critics of scientific values may have wider use in pointing to wider varieties of power discrepancies and other political- norrnative values within medical explanations. Instead, I think my project is complementary to many feminist critiques of medicine and of science, even if it is not itself a feminist project. A theory of pragmatic explanations (a la van Fraassen) can be feminist-friendly, that is, identify and address social inequalities within medical practices (specifically, the creation of explanations of disease). This complements an understanding of medical practice as seeking to improve patients’ lives; as part of this, most clinicians would be open to projects that can identify and reduce any harm that may come from standard medical practice and theories. Finally, I use these feminist values in conjunction with van Fraassen’s pragmatics of explanation to illuminate that medical explanations have social value, rather than purely scientific value. I have not argued in any strong sense for the appropriateness of these as feminist values; even Longino leaves this open to debate. Instead, I want to show what can be done when we seriously engage with such values as part of generating and evaluating clinical explanations. We can identify some of the work in explanations that is already going on (say, when patients are asking to be heard), as well as provide grounds for the critical and rigorous evaluation of medical explanations (and rejection of non-helpful or inaccurate explanations). A useful theory of medical explanations should address a wide range of social norms, without focusing exclusively upon women’s oppression. I think medical 203 explanations (and medical practice, in general) should take into account feminist theory, as well as discussions of racism, homophobia, sexism, ageism, class/economics, and other social factors. Of course, I do not mean to imply here that these ‘-isms’ are a package deal; there are occasionally conflicts when trying to promote, say, gendered and racial equality. My goal is to promote theorizing of social factors as important for the generation of medical explanations. Medical explanations can incorporate social determinants of health as explanatory, and consider feminist epistemic values, without sacrificing scientific rigor, without the explanatory process slipping into a purely normative or relativistic project. I do not defend this modification of our understanding of all scientific inquires in this chapter. Instead, I argue for looking at medicine, especially medical explanations, as necessarily a combination of science and ethics. I will leave the discussion of whether science can similarly be expanded to another paper or another author. How Does the Articulation of Values Improve Evaluation and Creation of Medical Explanations? As Longino notes, there is no single way to group or compare the elements of feminist and traditional scientific values. Here is one possible way of comparing these values, as provided by Longino:264 1 Feminist List Traditional List Empirical adequacy Accuracy Novelty Intemal/External Consistency Ontological heterogeneity Simplicity Complexity of interaction Breadth of scope Applicability to human needs Diffusion of power Fruitfulness 2"" Longino, “Gender, Politics, and the Theoretical Virtues”, 392. 204 I will be looking at certain of these pairings, and describing how I see these values as shaping medical explanations. I look at the values of empirical adequacy versus accuracy; of fruitfulness versus diffusion of power and applicability to human needs; of external consistency versus novelty; and of simplicity versus ontological heterogeneity. Other comparisons can be made, such as those provided at various points by Longino, but these few will suffice to make my points about how values (of both clinicians and patients) shape medical explanations. Longino’s values also provide means to better understand where scientific and "Q. medical explanations may differ. Below I show how standard versus feminist values shape the clinicians’ requirements on medical explanations. What work does Longino’s list of values do, work that Kuhn’s list (i.e., accuracy, fruitfulness) does not do? In some cases, Kuhn’s values may not speak to the needs of patients, or will mask some of the problems patients face during the explanatory process. Utilizing Longino’s values makes some of this work clearer. Medical explanations, as a practical exercise, are meant to benefit patients by improving their understanding (as well as describe how the world works). Thus, these feminist values work to better address patients’ explanatory needs without slipping into explanatory relativism. Accuracy & Empirical Adequacy Medical explanations require some reasonable requirement that the explanation adequately reflects the reality of the world. For this reason, Kuhn argues for the scientific value of ‘accuracy’; Longino argues for empirical adequacy.265 In generating medical 2“ Note that empirical adequacy is rooted in van Fraassen’s position as an anti-realist in The Scientific Image. He argues that empirical adequacy is the best that science can achieve, and that a deeper “accuracy” or “truth” is beyond the ability of science. In this way, empirical adequacy is already a component of the pragmatic explanations I argued for in the previous chapter. Empirical adequacy will not determine the members of the contrast class or which relevance relation is used. Instead, it is used in a means of determining whether there is good evidence to consider some causal factors as explanatorily relevant as compared to other causal factors. 205 explanations, clinicians’ scientific background will demand proof or evidence that a theory is correct. In some sense, calling a medical explanation ‘scientific’ often meanst it happens ‘not by chance’ or ‘has good supporting data that has not been falsified’. Although patients without a scientific background may not refer to it under these terms, it seems rational that patients will want medical explanations to match up with ‘the real world’, that is, theories ought to be (approximately) correct. Patients seek clinical knowledge and expertise, assuming it will be accurate. Part of clinicians’ expertise (and therefore the scientific accuracy of their explanations) is tied to their training in biomedical science. Accuracy here is tied to the accuracy of biomedical research. Yet clinicians are tied to accuracy in another way. Part of the clinical skills that clinicians develop over time—their bedside manner and their clinical skills—helps them to translate their scientific knowledge into better communication with patients. As their clinical skills improve, often through practice and repetition, clinicians become more accurate, that is, more skilled and nuanced in their ability to communicate effectively with patients. Patients seek the advice of clinicians as experts for both of these reasons. It should be noted, though, not all these scientific values (Kuhn’s or Longino’s) necessarily work well together. Even Kuhn’s accuracy may come into conflict with simplicity: the more accurate explanation may not be simple, and increasing simplicity may lead to errors or poor descriptions. How might the value of empirical adequacy be utilized when generating medical explanations? If there are two competing explanations, the one with higher empirical adequacy is favored more. As an example, patients may ask physicians what reasoning they have for, say, the recommendation of a colonoscopy (which patients often see as a rather invasive and intimidating procedure). The physician can point to data suggest that patients of a certain age (say 50 or older) ought to have regular screening for polyps that may lead to colon cancer; this regular screening may reduce chances of developing 206 cancer (or catch it before it spreads). This is information about the patient at a rather abstract level, where this patient is grouped among all persons over 50 (rather than just about this specific person). Once the test (here, the colonoscopy) is performed, the level of data about this specific patient increases. The doctor has better reason for assessing risk, for making suggestions for dietary changes or other medical changes for this patient. The test has provided better data for these suggestions. At various points, patients can reject the physicians’ theory or explanation, and provide their own theory. Here, too, valuing empirical adequacy can allow the physician to ask of the patient, “What is your reason for believing in that? What evidence do you have?” If the patient replies with information of how the stars control his blood pressure, the physician can ask for supporting data. In such a situation, the doctor can dismiss this information as being non-explanatory if the data contradicts the theory. If accuracy is a value to both patient and clinician, then there will be times when both parties need to decide how to act if an explanation is not strongly accurate. If a clinician is unsure about how accurate an explanation is, she should be upfront about this limitation.266 This provides patients other options about how to act: they can trust the doctor’s clinical judgment (making up for the lack of scientific accuracy) or on faith alone; or, patients can seek advice from different clinicians. Here, the point is that some work can be done, even when an accuracy criteria for explanations cannot be met: patients and clinicians can still have conversations about medical error, the lack of certainty about empirical support, how long the should wait until deciding how to act. 2“ This is not the only option, since the clinician could try to convince the patient by other means; but to do so seems dishonest from the beginning. 207 Fruitfulness versus Diffusion of Power and Applicability to Human Needs How can medical explanations be fruitful? That is, how does an explanation give rise to other explanations? One example comes from Thagard’s work where he discusses the historical changes in medical theories about ulcer formation. Ulcers had been thought to be caused by poor diet, or by stress, which each led to higher stomach acid levels, which caused the ulcer. Currently accepted theories today say that bacterial infections are the cause of ulcer formation. In this case, germ theory is a background explanation of other phenomena. This explanation led to the discovery of bacteria, viruses, and other infectious microbes. This in turn set the stage for the explanation that bacterial infections are explanatory for the causation of ulcers. Here, the discovery of one biological theory (germ theory) became explanatorily fruitful in that it allowed the formation of a number of diverse and separate explanations. Without germ theory, it is likely that the bacterial infections of ulcers would never have been discovered, and thus alternate (and probably poorer) explanations would be used instead in ulcer treatment. Let us consider a different example of explanations as fruitful, one involving a specific patient. A woman speaks with her physician because she is having problems sleeping. She feels fatigued, even after waking in the morning, and is snoring. She is diagnosed with episodic oxygen desaturation, that is, low blood-oxygen levels when sleeping. The doctor uses this diagnosis as an explanation for blood pressure and fatigue. But, upon listening to the doctor, the patient finds this diagnosis also explains other symptoms she had previously thought unrelated, and thus had not brought up to the doctor. She has been suffering from headaches, which she thought was caused by job stress. She had also experienced a strange incident where she woke up in the middle of the night, but was unable to move. This sleep-paralysis and the headaches, things she had not thought to bring up to the doctor before, were symptoms the doctor said other patients with episodic oxygen desaturation sometimes report. In this case, this diagnosis is 208 ‘fruitful’ in that it explains more events in the patient’s case than she had originally sought to have explained. How is this value of ‘fruitfulness’ compared with Longino’s two feminist pragmatic virtues of applicability to human needs and of distribution of power? I will begin by looking at pragmatic medical explanations and the value of applicability to human needs. We might begin by questioning the applicability of this virtue to medicine. Even if we grant the Longino has argued convincingly for the role of this virtue in scientific practices and research, does it have a place in medical explanations? If we begin by asking, “are medical explanations applicable to human needs? ” the apparent first response might be, yes, medical explanations must be applicable to human needs since this one branch of scientific research, among all others, is devoted to human well- being. But if this is the case, this limiting virtue does no work for medical explanations; it would automatically cover all medical explanations. Instead, there may be stronger and weaker degrees of applicability to human needs as a condition for medical explanations. An example of a weaker degree of applicability to human needs might be projects that have no immediate or obvious connection to improving human health. In some senses, it was unclear whether early research into the Human Genome Project would produce useful information that may help actual patients. Some critics argued it was a biological experiment that was “for knowledge sake alone”, not something that would help actual patients. As it turns out, there has been some benefit to patients from results of the HGP, even if advanced activities like gene replacement therapies are not yet safe enough for wide-scale human use. Another example of a weak sense of applicability to human needs might be continued research being done on smallpox. This disease has been eliminated in natural infections due to the World Health Organization campaigns, which involved a rather successful vaccine and vaccination strategy (although some persons did become infected from the vaccine itself). The virus exists now only in labs. A case could be made that continued research on this disease does not meet the criterion of 209 applicability to human needs, since no humans are suffering from smallpox; as such, no patients will benefit from improved explanations of this disease that may result from more lab research. At best, further research can only meet the needs of possible future persons, say, if the virus was used as a biological weapon. But to continue such research means making choices that prioritize the distribution of limited resources—time, money, and scientific interests—on a project with hypothetical advantage (treating or preventing future smallpox cases) over projects with current human needs (hunger, cancer, HIV/AIDS research). As such, further research on small pox may fail to meet the criterion of applicability to human needs. A stronger degree of applicability to human needs may be the following example. As described by Kuhn, fruitfulness as a virtue focuses explanatory attention inward (further within the theory). Thus, explanations that implement this virtue generate other explanatory projects, questions, and answers. What is not brought to light by fruitfulness is a greater understanding of how the explanation generates or sheds light on problems or solutions for the lived experiences of patients. In the case of this sleep disorder, the fruitfulness of an explanation is a function of its ability to account for other symptoms the patient may suffer from (e. g., headaches, groggy feeling, etc). Applicability to human needs (and maybe diffusion of power, discussed below) is not a value in each and every explanation dust as the fruitfulness of a theory may be great in one context but poor in another). Taking applicability to human needs and diffusion of power seriously “requires looking beyond the immediate (internal) context of research to the ways in which that research might or might not be developed. This in turn requires taking stock of the social, political, and economic context in which development might take place.”267 This means turning our explanatory gaze ‘outward’ to understand the context and forces that explanatory production, and the effects of this 2‘” Longino, “Cognitive and Non-Cognitive Values in Science”, 53-54. 210 explanation on patients and clinicians. This may be how many other patients’ experiences can be similarly explained, or if it looks at other issues besides the immediate case (e. g., maybe it adds to understanding about other illnesses brought about by low oxygen levels). In the sleep disorder case discussed earlier, applicability to human needs works to focus our attention outward from the explanation, showing us that there may be other things involved: what forces in her life prevent her from being active or from losing weight? Is she providing care for others? Is she feeling unsafe such that she avoids exercise at night? Has she been trained against seeing her body as ‘active’ such that she has never developed a habit of sports and physical activity? In this case, these factors shape the explanation by changing the relevance of certain information. If the patient is unable to understand (too complex) or unable to implement certain changes (her fear of falling keeps her from exercising), then some explanations will not be helpful to her, even if they are ‘correct’ explanations. Accurate molecular explanations may not answer a patient’s questions (relieve her puzzlement) because this is not something she can work with. But, if an alternate strategy for explanation is available to the clinician, the value of ‘applicability to human needs’ may compel the clinician to rework the explanation. Had the patient not brought up further information, the explanation would have had a smaller scope. Had she been too busy to seek the doctor’s advice, the explanation may never have been generated. As such, these are all factors shaping the medical explanation, and most likely shaping those of other female patients with similar lives. Notice that the value of applicability to human needs has a connection with the way in which why-questions are interpreted and responded to. Rather than understanding medical explanations as scientific accomplishments, I have argued they should primarily be understood as tools to improve patients’ understanding. I have thus argued for understanding them as erotetic explanations, rather than ontic explanations. But in a 211 richer sense, applicability to human needs asks us to examine the structure of individual explanations more closely. By understanding how relevance relations are implied or intended by patients, and how these may differ from those of clinicians, we are reminded that different parties may have different intended meanings to their why-questions. As I argued in Chapter 5, in some cases clinicians should try to utilize relevance relations that are not what they may be accustomed to, in order to understand and provide the explanations that patients may want—explanations that may be utilizing different relevance relations. It is this search to identify and utilize different relevance relations that may allow research explanations to be changed to clinical explanations—that is, by identifying the explanatory needs of patients, we can make medical explanations more relevant to their current needs. Fruitfulness as an explanatory value can also be contrasted with diffusion of power. Diffusion of power is what Longino calls the practical version of mutuality of interaction (or ontological heterogeneity). This is a different strategy for taking our explanatory gaze outward (rather than inward, as with fruitfulness), so that we can better understand the additional factors influencing medical explanations. This value allows for the political and moral work of explanations to be revealed, by showing the forces at work shaping the process around generating explanations. Historically, physicians and other medical clinicians have been viewed as experts, and often as experts whose decisions should not be questioned. Much of basic bioethics has been to challenge the authority implicit to such a relationship and to seek means to empower patients. Relatedly, the feminist value of diffusion of power works to examine the power dynamics implicit in the clinician-patient relationship, since this is the setting where medical explanations are generated. What would it be like to enact diffusion of power in medical explanations? One sense is to see both the clinician and the patient as epistemically valuable. In other words, one way to implement this value is to see both clinicians and patients as epistemic agents 212 who can add useful information and together create a medical explanation. Both groups can add to the process of generating explanations, especially in how interests shape/limit the structure of contrast classes and relevance relations. Diffusion of power also requires that both parties be heard and taken seriously, not just that they have the potential to be epistemic agents. So, there is a mutual obligation between physicians and patients. Physicians have historically failed to meet this criterion by seeing the clinicians’ role as the decision-maker, and dismissing the patient as uninterested in their own health care or lacking sufficient skills (overall interest, intelligence, specific) necessary for deliberation as part of health care decisions.268 A critical clinician, though, may argue that diffusion of power is an impediment to medical explanations because it would require too much. Taken strongly, some may see this value as requiring a clinician to “teach the patient everything I learned in medical school”, as is often a criticism discussed in issues of informed consent. But I see this response as an unwarranted criticism.269 Such critics often say to really get informed consent, you would have to have the patient go through years of medical training before she could fully understand what is being discussed. If patients were to take part in medical explanations, they would (similarly) need significantly more medical knowledge before this practice would be meaningful. In fact, it is exactly this medical knowledge that brings a patient to a clinician; if they were peers, the patient would not need the advice of the doctor.270 268 Here, I am thinking about the clinical role as described and criticized by Jay Katz in The Silent World of Doctor and Patient. 26” I am generally suspicious of this line of thought because I think this is often delivered not as an argument (even as a poorly framed argument). Instead, I think it is a fallacious argument meant to be a conversation-stopper. Such a response ignores the idea that part of medical expertise may involve developing the proper communication skills to pass along information to patients (on the right topics and at the right level of scientific detail). 27° Anyhow, as Robert Veach says about the role of lay-persons in medicine, “it is quite reasonable to supply an adequate lay assessment of the medical facts. It is done every time a lay person uses an over-the- counter medication, tries a home remedy, or decides against seeking professional medical advice for a 213 Yet to take this line of argument is to misunderstand diffusion of power, or to wrongly assume it means equality within the clinic. Diffusion of power does not require equality of power (or knowledge). Diffusion of power does not assume that each of the epistemic agents is “granted equal authority on every matter.”271 Instead, valuing diffusion of power means we should favor theories that improve patients’ (especially women’s) lives without using or relying upon structures/pattems of inequalities of power. Medical explanatory theories that see the clinician as the primary (or only) epistemic agent are examples of theories that rely upon an inequality of power: the clinician’s power and authority is maintained by the role of explainer. This role is one where clinical judgment cannot be questioned, since the patient is seen as being without the tools necessary for meaningful criticism. A pragmatic explanatory model instead diffuses this power by sharing epistemic work between clinicians and patients (and possibly others). As such, a pragmatic model of explanation works to empower patients, rather than maintain patients’ ignorance or undermine their ability. To borrow a term, pragmatic medical explanations have the goal of “transparency”.272 One way in which clinicians can empower patients through improving their understanding is to make the explanatory process clear, visible, and understood by patients. When patients are seen as part of the explanatory process, it will be less likely that explanations will be used against their interests. In other cases, it may ensure the explanation addresses the patients’ needs, rather than focusing deeper into the fruitfulness of the explanation, as was discussed above. In the previous chapter, I noted that in the back-and-forth conversations between clinicians and patients, explanatory conversations often occur. Explanatory requests may be edited, or occasionally rejected. Recall the example of the patient, Paul, who is HIV +. He asks why is he sick, but his problem. Lay people. . .will almost always supply a large portion of the facts for what are generally thought of as medical decisions” (“Lay Medical Ethics”, Journal of Medicine and Philosophy, 10 (1985) 3). 27’ Longino, “Interpretation Versus Explanation in the Critique of Science”, 1 19. 272 Howard Brody, “Transparency: Informed Consent in Primary Care”, The Hastings Center Report 19 (1989): 5-9. 214 partner, Nick, remains HIV -. If the why-question is stated as, “why is {Paul, Nick} sick?”, I argued that this is an improper use of contrast classes. It assumes that one and only one of these people can become ill, which is incorrect. The clinician in this case can teach the patient to sort through his explanatory requests: “Why is it the case that Nick {is, is not} sick?” and “Why is it the case that Paul {is, is not} sick?” The utility of diffusion of power in this case involves recognizing that patients can be taught how to ask better why-questions. A second critic might ask about the patient who is non-compliant, or who does not want to be involved in the explanation generating. Certainly some patients will not want to take the time and energy to participate this actively since physicians often encounter patients who do not want to participate in decision-making regarding their own health care; this may be due to any number of factors. Other patients may be unreflective about their choices. Some patients may want clinicians to decide for them; some patients may be so openly hostile to clinicians that the interactive nature of pragmatic explanations is impossible. Does the well-meaning doctor (one who is savvy about social power dynamics in the clinic) need to somehow ensure (forcefully, perhaps) diffusion of power? That is, does a pragmatic model of explanation require the patient to take part in generating explanations? If a goal of pragmatic explanations is to empower patients, then no, forcing patients to participate against their will is not called for. Such force actively works against patients’ empowerment by taking away their choice. This critic’s move also misunderstands what is at the heart of this value by seeing only a single way of patient empowerment. External Consistency versus Novelty Internal consistency seems easily accepted. It is easy to understand why explanations should have internal (logical) consistency. This is not exclusively a 215 scientific requirement, since non-scientists will still require that explanations “make sense” or “seem logical.” The value of external consistency is favored in standard medical explanations to limit non-productive information from entering into explanations. For instance, Salmon and Kitcher’s worry regarding the explanatory value of astrological data may be in large part a call for external consistency: astrological theories do not mesh with other scientific theories, (scientific/medical) explanations should not rely upon astrological data. Valuing external consistency keeps inappropriate considerations (non-medical, or non-scientific) out of our explanations, and thus increases the probability that the explanations are true. Yet a strong reliance on external consistency (as opposed to the feminist value of novelty) may result in different problems. Novelty is embodied in theories “that differ in significant ways from presently accepted theories, either by postulating different entities and processes, adopting different principles of explanation, incorporating alternative metaphors, or by attempting to describe and explain phenomena that have not previously been the subject of scientific investigation.”273 So, external consistency, when taken strongly, can be akin to thinking in certain patterns or ruts, which may negatively affect patients in how explanations are generated. Novelty is the expressed interest of thinking in ways to avoid entrenched patterns.274 One ‘rut’ involving a medical theory is the concept of HIV/AIDS as a “gay disease.” Michael Scarce provides an enlightening history for why HIV/AIDS was readily adopted (by clinical and popular cultures) as a disease specific to (or predominating in) gay male communities. Scarce details a predecessor disease, that of Gay Bowel Syndrome (GayBS). In 1976, Henry L. Kazal and his four co-authors 273 Longino, “Cognitive and Non-Cognitive Values in Science”, 45. 27" Although Longino does not explicitly cite this, her argument about theory acceptance and change is similar to those of Paul Feyerabend, “Explanation, Reduction, and Empiricism” in H. Feigl and G. Maxwell, (eds), Minnesota Studies in the Philosophy of Science, 3, (1962): 28-97. 216 published the first clinical article describing “Gay Bowel Syndrome.”275 Gay Bowel Syndrome is described as “a pattern of sexually transmitted anorectal and colonic diseases encountered in 260 predominantly white, middle- to upper-class, gay men who had visited a private proctologic practice in New York City.”276 GayBS is not a single condition, but a pattern of multiple possible conditions that seemed to have occurred with unusually higher frequency with a specific population, that of gay men. Yet diagnosis was problematic, even from the conception of this disease. Notice Scarce’s description of diagnosis of GayBS, so complex and disconnected as to almost be ludicrous: Clinical diagnoses of these patients included twenty-two conditions (in descending order of prevalence): condyloma acuminata (genital warts), hemorrhoids, nonspecific proctitis, anal fistula, perirectal abscess, anal fissure, amebiasis, pruritus ani, polyps (benign), hepatitis, rectal dyspareunia, gonorrhea, syphilis, trauma and foreign bodies, shigellosis, rectal ulcers, lymphogranuloma venereum, anal incontinence, solitary rectal ulcer, Bowen’s disease, squamous cell carcinoma, and an ‘other’ classification.277 Diagnosis of GayBS depends on identifying one or more of these conditions as present in a patient. Over the following years, GayBS has become entrenched in medical practice and theory. Rather than becoming more concise, though, the list of conditions associated with GayBS has grown into a list of over 50 items. The definition and research of ‘gay bowel syndrome’ was problematic in various ways from the beginning use of this term. First, the physicians who authored the original study on GayBS failed to obtain a complete sexual history of the men. In fact, it was not clear that each of the men actually had engaged in homosexual sexual activity. This failure to include patients’ sexual history generated incomplete explanations of this disease. There were assumptions made about who was gay, and what that label indicated 2” From Annals of Clinical and Laboratory Science, as cited in Michael Scarce Smearing the Queer: Medical Bias in the Health Care of Gay Men (New York: Haworth Press, 1999): 14. 276 Scarce, Smearing the Queer 14. 277 Scarce, Smearing the Queer 14. 217 (i.e., anal sex assumed, as well as sexual/physical contact assumed). This was the result of physicians being allowed to determine (as medical lscientific experts) what the important features of the explanation were going to be. The label “gay bowel syndrome” (and the underlying explanation) fails to be complete. How does GayBS, even if it is a flawed disease concept, relate to HIV/AIDS and the concept of external consistency? Discussions of HIV /AIDS in the early 1980s maintained some of the major metaphors, assumptions, and theories set in progress by GayBS. This, in and of itself, is not unique to HIV. Medicine does it all the time, and perhaps this strategy works in many cases, and as such it preserves the value of external consistency. In this case, though, concepts such as the “fragile anus” and “rugged vagina” explained why HIV/AIDS was a gay disease; certain body parts, when abused through homosexual (or unnatural) sexual behavior resulted in infection. Viewing AIDS as a “gay” disease fit neatly with other theories, about the unnaturalness of homosexuality and scientists’ heterocentric views of sexuality. Yet valuing external consistency worked against proper understanding of HIV. The concept of a disease being ‘gay’ or ‘gay-related’ persisted, despite early data to the contrary. Even in the first epidemiological reporting of HIV, many men described were gay, but there were also a number of straight men, too. Despite this fact, HIV/AIDS was quickly taken up in the literature as ‘gay related immunodeficiency’ (GRID). Even after being renamed AIDS, and the clear discovery of HIV, the disease remained entrenched as a (second) gay disease. The value of novelty need not work in an extreme manner in order to be useful to medical explanations. In this case, novelty does not require that we discard past medical discoveries and advances. In terms of understanding how this can be seen in a pragmatic medical explanation, it may be that consistency and novelty work at the level of contrast classes. If the question being explored is, “What causes this disease?”, the contrast class might be something like {a virus; (immoral/unnatural) homosexual behavior; chemical 218 exposure}. The legacy of GayBS was that medical researchers held too strongly to the idea of ‘homosexual’ behavior being involved, rather than sexual behavior (generally) or a sexually transmitted pathogen, like herpes or syphilis. In this case, knowledge of viruses was central to identifying the infectious agent and seeking treatment options. What could have been avoided, though, was the unreflective incorporation of certain metaphors and patterns of thinking that proved to be distracting to early HIV research as well as promoting further homophobic discriminatory practices within medicine. Simplicity versus Ontological Heterogeneity As I described earlier, the Kuhnian value of simplicity states that when presented with two explanations of the same event, simplicity favors the explanation with fewer complex mechanisms or assumptions. This value is in contrast with Longino’s ontological heterogeneity, which values the consideration of a variety of causal events and pathways. This valuing of ontological heterogeneity, Nelson notes, may occur in two different ways. First, feminist science writings have often argued for being able to account for individual differences (why one member is different from others within a sample population). Barbara McClintock’s work on corn kernel variations here is often cited as the paradigmatic example of such work.278 Her detailed attention to the variation among corn kernels on a single ear helped her “to recognize an underlying pattern of mutability.”279 In some ways, this may be in contrast to Kitcher’s unificationist theory: explanatory projects focusing on the individual—rather than wide patterns and generalizations—are also of value. 273 E. F. Keller, A Feeling for the Organism: The Life and Work of Barbara McClintock (San Francisco: W.H. Freeman, 1983). 27” Longino, “In Search of Feminist Epistemology.” 219 ". The second way in which ontological heterogeneity is valued is that it supports arguments that “difference is a resource, not a failure.”280 Nelson notes that standard science has prioritized difference: one type is chosen as the standard, while other types are failed or incomplete versions. Yet by embracing difference, multiple types are allowed equal standing, and thus can be investigated with equal worth. In this way, ontological heterogeneity as a feminist value complements nicely my .1 previous arguments that (a) social determinants of disease must be considered as explanatory, and (b) Thagard’s CN I is deficient since it cannot account for social forces as explanatory of disease. Ontological heterogeneity as a value incorporated into clinical { explanations will favor explanatory strategies that allow for consideration of social forces over theories that cannot. First, ontological heterogeneity allows for explaining individual cases as valuable. When such explanations do not fit within the larger pattern, it allows for a more detailed examination of the situation. In the cases of Uncle Bill and Julia, rather than focusing on how they fit into the explanatory patterns of persons with cancer or ulcers, clinicians would be better served by understanding how their cases are different from the paradigmatic cases. In these cases, this will involve in large part understanding the influence of social forces on these patients’ health. Here, difference is not a failure of explanation (failure to fit into paradigmatic situations); instead, difference allows a more detailed consideration of what medicine has often overlooked (the social determinants of health. One might notice, though, that complexity plays a role both in Thagard’s CNI and in Longino’s ontological heterogeneity. Although Thagard’s theory does not account well for social forces, many CNI are highly detailed and intricately complex. In fact, in Chapter 3 I criticize Thagard’s CNI for being overly complex. Are the examples of 28° Longino, “In Search of Feminist Epistemology.” 220 Thagard’s CN I utilizing Longino’s value of ontological heterogeneity? I would argue that they are not for two reasons. First, although CN I may often begin as complex webs, they are often of certain types of information. Thagard argues for the importance of certain types of information as explanatory (e.g., genetics, microbial, nutritional information) but ignores other information (e. g., social forces). In this way, he incorporates a variety of information, but his model is not open to considering radically new types of causal information. Simplicity may prevent us from understanding the complexity of differently situated experiences. As such, people’s experiences may be dismissed because they do not live up to a given standard, rather than being incorrect or not meeting empirical adequacy. Second, Thagard’s CNI often begin as highly complex webs, which I compared to Railton’s ideal explanations. CN I are generally about types of disease cases. Yet when generating clinical explanations, I have argued in Chapter 3 that the web is often “whittled down” to the explanatory thread that best describes the case at hand by eliminating possible causes that are not involved in this case. The goal for Thagard’s version of clinical explanations is to find the single causal thread that is relevant to case at hand. In generating clinical explanations from CN I, the goal is always to move towards more simplistic explanations. In some cases, though, it will be explanatorily important to consider additional possible causes, rather than minimize the possibilities. That is, in some cases, we will want to move away from simplicity as a guiding value. Consider the case of J.S., a woman suffering from what has been called “unexplained medical disorders.” A clinician generating a CNI for this patient will work to find the single causal thread that is resulting in J .S.’s ill health. The problem, though, is that different clinicians have generated numerous explanations, each citing different causal factors. J .S. has received a number of different, often conflicting, explanations. J.S. begins to feel frustrated with her doctors. 221 Ontological heterogeneity as an explanatory value, though, helps to avoid this problem. As I argued in Chapter 5, it is possible to have different relevance relations for different explanations of this case. Some clinicians may seek physical causes, while others focus on psychological causes. Still, other clinicians may try to understand the explanatory role of seemingly unrelated factors of the patients past. Consider the fact women with unexplained medical disorders are more likely to have histories of sexual assault. Typically, the women have seemingly recovered (physically and/or psychologically) from the rape, and hence many clinicians dismiss the incident as explanatory of current health problems. It is interesting to think about why such causal threads are often disregarded by clinicians. Likely the reasons vary. Yet some light on this is shed by a Canadian study of clinicians’ inquiries with patients with irritable bowel syndrome (one of the unexplained medical disorders). While the study found that 90% of clinicians thought routine inquiries into physical or sexual abuse should be made, and only 10% of doctors found such information to be “irrelevant to management,” only 23% always or frequently collected such data. Clinicians reported other factors for why they did not collect information on sexual and physical abuse, such as time constraints, and lack of resources for treatment referral (“If we can’t treat it, we should ignore it.”) Yet I agree with the authors of the study who write, “Patients have the right to understand causal or contributing factors to their disorders irrespective of treatment options.” As they rightly point out, numerous patients are informed of organ failure, even if their chances of receiving an organ transplant are minimal to nonexistent?81 In the general case of unexplained medical disorders and the specific case of J .S., ontological heterogeneity favors keeping multiple explanatory paths open, rather than 28‘ Ilnyckyj and Bernstein, “Sexual Abuse in Irritable Bowel Syndrome”. The authors continue, noting, “in the case of abuse. patients may be encouraged by the link to their illness to help themselves or to seek help outside the health sector.” 222 reducing the case to one explanatory pathway.282 Ontological heterogeneity helps to keep explanatory options open, rather than to seek explanations that are simplistic but inadequate to the case. Conclusion In this chapter, I have argued for understanding a wider range of values as influencing explanation production. By drawing upon feminist critics of science (mainly the work of Helen Longino), I have argued for taking into account constitutive values that also have political-normative interests (rather than those with only epistemic interests). These values are beneficial to a theory of pragmatic explanations in two ways. First, they better reflect the actual practice of medical explanations. Medicine has both epistemic and social goals. Standard models of medical explanations fail to account for the social- political values within medical explanations. Second, these feminist values work well to empower patients and prevent further oppressive processes within medical science and practice. If medicine seeks to improve patients’ lives, then a system of research that itself causes harm must be critically examined and improved. Incorporating feminist values within pragmatic medical explanations is one step towards this goal. 282 When clinicians attempt this reduction, often women feel frustrated with the answers (e. g., it’s not obviously physical, but I’m not mentally ill. . .). 223 CHAPTER 7: CONCLUDING REMARKS The topic of medical explanations of disease is challenging, fruitful, and important for a number of reasons. First, as I have argued, medical explanations, especially clinical explanations, usefully highlight explanatory projects as having both epistemological and moral/political aspects, which are intricately connected and raise important questions. For instance, I have criticized the standard account of the explanatory agent—the “view from nowhere”—as improper for clinical explanations. Clinical explanations typically involve at least a patient and a clinician, yet in addition others may be involved in the explanatory process, e.g., a range of physicians, nurses, family members, counselors, and clergy. Understanding the social situatedness of the explanatory agents becomes important in correctly identifying the explanatory context from which why- questions are asked. To take a specific example, I have noted the complexities involving gender and explanation. Rather than disembodied agents, clinical explanations are requested by lived beings from specific social positions. It may not be surprising to some that “medically unexplained disorders” are highly gendered. I have noted a number of possibilities for this, but two likely contributing factors stand out. First, women’s voices, and thus their explanatory requests, may often go unheard by the medical profession. Second, the women’s health movement has continued to argue that women’s health issues often go under-researched within medicine. So even if an explanatory request is “heard”, there may be little relevant science on which to base an explanatory response. The point to be raised here, though, is that social situatedness within explanations needs further examination. For instance, I believe important work remains to be done regarding the connections between race, culture, and explanatory requests. This may involve questions 224 of whether race shapes explanatory requests in important ways, or whether race influences—that is, hinders or promotes—the success of explanatory responses. My work on clinical explanations also has implications for the evolution of the doctor-patient relationships. I have argued that patients should be taken seriously as epistemic agents. Yet we must ask, how will this affect how patients relate to clinicians? One possible response that warrants further examination is whether increasing patients’ participation in clinical explanatory projects subsequently increases their personal responsibility. Asked another way, what moral responsibility do patients and clinicians have as epistemic agents to engage well in clinical explanations: not just to bring about improved understandings, but also to articulate how people are set in specific social relationships to one another. The concept of epistemic responsibility is one that warrants further philosophical research. I believe the interesting work of social, feminist, and medical epistemology will continue these conversations in the years to come. My hope is not to provide an account of medical explanations that requires the doctor-patient relationship to be viewed as symmetrical or as one between “equals”. Instead, articulating the roles of the various explanatory agents allows for self-reflection, collaboration, and, perhaps most importantly, continued conversations about the patient’ 3 health. The ultimate goal—tire improvement of patient healthcare—should not be overlooked. Empirical research regarding clinical explanations would be useful here on two points. First, does a pragmatic account of explanation actually facilitate conversations between patients and clinicians? Second, does an improved account of clinical explanation result in improved health care services for patients? These empirical considerations are important. From a philosopher’s perspective, a pragmatic account of clinical explanations may be superior when judged against other options because it describes the nature of the world more accurately. Yet it may be difficult to incorporate my account of clinical explanations into current medical practice without proof of some benefit. Habits and clinical practice are slow to change. If it could be shown that a 225 pragmatic account of clinical explanation—one which considers the epistemic role of patients seriously and thoughtfully—improves health care outcomes, there is greater likelihood for clinical uptake and change. If both clinicians and patients believe that acting differently will result in a true benefit, they may be more open to the challenges of releaming their roles in the doctor-patient relationship. While these may be clinically important implications of this work, the topic of clinical explanations raises important considerations for explanation theory within medicine, and science more generally. In advancing the work of van Fraassen, I have argued that explanations are necessarily contextual. In this way, Salmon’s and Railton’s acontextual approaches to explanation, while thought provoking, are of limited utility. In my work 1 have raised questions about what the nature of explanatory context involves, yet more work remains. For instance, both Salmon and Railton hint that the project of articulating explanatory contexts will be complicated and difficult. But the lived practice of clinical explanations shows us it is possible to have such conversations, and that the people involved may appreciate guidance. On a different point, Salmon and Railton’s discussions of scientific explanations have focused on what I have called peer-peer explanations; that is, explanations shared between people with roughly similar scientific backgrounds. This discussion of clinical explanations calls to our attention that explanations about medicine often occur between non-peers. This complicates the contextuality involved in pragmatic explanations beyond what even van Fraassen discusses in his original work. Finally, I have raised a set of important questions about a pragmatic account of clinical explanations regarding the interplay of context and of the agents of clinical explanations. I have argued that the solipsistic individual is inadequate for clinical explanations, and I instead argued for taking on Lynn Nelson’s epistemological community as the primary agent. In Chapter 4, I focused primarily upon the implications of viewing the doctor and the patient as an epistemological team that asks why-questions 226 about patients’ health. Clearly this is only one of many possible communities that may make explanatory requests, and each community will raise different questions about explanatory contexts. For instance, the sense of “community” becomes further complicated when thinking about medical explanatory request within public health. Public health programs study the health needs of a given region, e.g., the urban neighborhoods of a mid-sized city. Within that region, there will likely be a diverse population with possibly conflicting sets of needs. In such an example, Salmon’s and Railton’s worries seem reasonable: identifying explanatory contexts seems to be a daunting task. Instead, I view this example as a call for further philosophical research on explanatory contexts. Health information must be distributed to large, diverse populations. Although these projects may be complicated, they are important. To set such projects aside because of their complexity would be to make two mistakes: to continue explanatory practices that ignore socially relevant goals, and to fail to engage with and correct the limitations of current explanatory theories. 227 BIBLIOGRAPHY Achinstein, Peter. The Nature of Explanation. New York: Oxford University Press, 1983. Alcoff, Linda and Elizabeth Potter, eds. Feminist Epistemologies. New York: Routledge, 1993. Andre, Judith. Bioethics as Practice. Chapel Hill: University of North Carolina Press, 2002. Aronowitz, Robert A. Making Sense of Illness: Science, Society, and Disease. New York: Cambridge University Press, 1998. Benjamin, Martin and Joy Curtis. Ethics in Nursing. 2nd edition. New York: Oxford University Press, 1992. Boyd, Richard. “Metaphore and Theory Change: What is ‘Metaphor’ a Metaphor For?” In Ortony, 356-408. Brody, Howard. “Transparency: Informed Consent in Primary Care.” The Hastings Center Report 19 (1989): 5-9. Buckman, Robert. How to Break Bad News. Baltimore: Johns Hopkins University Press, 1992. Code, Lorraine. Epistemic Responsibility. Hanover, CT: University Press of New England, 1987. Collins, Patricia Hill. “The Social Construction of Black Feminist Thought.” Signs 14. 4 (1989): 745-773. Curd, Martin and J. A. Cover, eds. Philosophy of Science. New York: Norton and Company, 1998. 228 Cutter, Mary Ann G. “Explaining AIDS: A Case Study” in Eric T. Juengst and Barbara A. Koening, eds. The Meaning of AIDS: Implications of Medical Science, Clinical Practice, and Public Health Policy. New York: Praeger Publishers, 1989. .Reframing Disease Contextually. Netherlands: Klewer, 2003. Donald, Anna. “Balancing the Benefits and Harms of Care.” British Medical Journal 329 (2004): 59. Dresser, Rebecca. “What Bioethics Can Learn from the Women’s Health Movement.” Susan M. Wolf, ed. Feminism and Bioethics: Beyond Reproduction. New York: Oxford University Press, 1996: 144-159. . When Science Oflers Salvation: Patient Advocacy and Research Ethics. Oxford University Press, New York, 2001. Dretske, Fred. “Contrastive Statements.” Philosophical Review 81 (1972) 411-437. Dula, Annette, and Sara Goering, eds. “It Just Ain ’t F air”: The Ethics of Health Care for African Americans. Westport, CT: Praeger Paperback, 1994. Feyerabend, Paul. “Explanation, Reduction, and Empiricism.” In H. Feigl and G. Maxwell, eds. Minnesota Studies in the Philosophy of Science, 3, (1962): 28-97. Friedman, Michael. “Explanation and Scientific Understanding” Journal of Philosophy 71 (1974): 5-19. Frye, Marilyn. The Politics of Reality. Freedom, CA: The Crossing Press, 1983. Garf'mkel, Alan. Forms of Explanation: Rethinking the Questions in Social Theory. New Haven: Yale University Press, 1981. Gifford, Fred. “Genetic Traits.” Biology and Philosophy 5 (1990): 327-347. . “Explanations of Difference” (unpublished manuscript). 229 Grasswick, Heidi E. “Individuals-in-Communities: The Search for a Feminist Model of Epistemic Subjects.” Hypatia l9. 3 (2004) 85-120. Greenblatt, Samuel H. “Limits of Knowledge and Knowledge of Limits: An Essay of Clinical Judgment.” The Journal of Medicine and Philosophy 5. 1 (1980): 22-29. Harding, Sandra. The Science Question in F eminisrn. Ithaca: Cornell University Press, 1986. . Whose Science? Whose Knowledge?: Thinking From Women ’s Lives. Ithaca: Cornell University Press, 1991. . “Rethinking Standpoint Epistemology: What Is Strong Objectivity?” in Alcoff and Potter, 49-82. Hardwig, John. “Toward an Ethics of Expertise.” In Daniel Wueste, ed. Professional Ethics and Social Responsibility. Maryland: Rowman & Littlefield, 1994: 83-101. Hempel, Carl and Paul Oppenheim, “Studies in the Logic of Explanation.” In Hempel, 1965, 245-295. Hempel, Carl. “Aspects of Scientific Explanation.” In Carl Hempel, Aspects of Scientific Explanation and Other Essays in the Philosophy of Science. New York: Free Press, 1965, 331-496. Ilnyckyj, Alexandra and Charles N. Bernstein. “Sexual Abuse in Irritable Bowel Syndrome: To Ask or Not to Ask—That is the Question.” Canadian Journal of Gastroenterology, 16 (2002): 801-805. Johnson, L. Syd M. “Review,” http://mentalhelp.net/poc/view_doc.php?id=2066&type=book&cn=72, accessed September 9, 2005. Katz, Jay. The Silent World of Doctor and Patient. New York: Free Press, 1984. Keller, Evelyn Fox. A Feeling for the Organism: The Life and Work of Barbara McClintock. San Francisco: W.H. Freeman, 1983. 230 . Reflections on Gender and Science. New Haven: Yale University Pres, 1985. Kim, J aegwon. Supervenience and Mind: Selected Philosophical Essays. Cambridge: Cambridge University Press, 1993. Kitcher, Philip and Wesley C. Salmon. “Van Fraassen on Explanation.” The Journal of Philosophy (1987): 315-330. . Salmon, Scientific Explanation. Mirmeapolis: University of Minnesota Press, 1989. Kitcher, Philip. “Explanatory Unification.” Philosophy of Science 48 (1981): 507-531. . “Explanatory Unification and the Causal Structure of the World.” In Philip Kitcher and Wesley C. Salmon, Scientific Explanation. Minneapolis: University of Minnesota Press, 1989: 410-505. . The Advancement of Science. New York: Oxford University Press, 1993. Kuhn, Thomas S. The Structure of Scientific Revolutions Chicago: University of Chicago Press, 1962. . “Metaphor in Science” in Ortony, 403—19. Laudan, Larry. Progress and Its Problems: Towards a Theory of Scientific Growth. Berkeley; University of California Press, 1977. Lewontin, Richard C. Biology as Ideology: The Doctrine of DNA. (New York: Harper Perennial, 1993. Longino, Helen. Science as Social Knowledge. Princeton: Princeton University Press, 1990. . “In Search of Feminist Epistemology.” Monist, 77 (1994): 472-486. . “Cognitive and Non-Cognitive Values in Science: Rethinking the Dichotomy.”, In Nelson, L.H. and Nelson, J ., 39-58. 231 . “Gender, Politics, and the Theoretical Virtues.” Synthese 104 (1995): 383-397. . “Gender, Politics, and the Theoretical Virtues.” Synthese 104 (1995): 383-397. . “Interpretation Versus Explanation in the Critique of Science.” Science in Context 10. 1 (1997): 113-128. Mackie, J. L. “Causes and Conditionals.” American Philosophical Quarterly 2 (1965): 245-65. Malterud, Kirsti, Lucy Candib, and Lorraine Code. “Responsible and Responsive Knowing in Medical Diagnosis: The Medical Gaze Revisited.” NORA: Nordic Journal of Women ’s Studies 12. l (2004): 8-19. Malterud, Kirsti. “Symptoms as a Source of Medical Knowledge: Understanding Medically Unexplained Disorders in Women.” Family Medicine 32 (2000): 603- 1 l. Marmot, M. G. and George Davey Smith, et a1. “Health Inequalities Among British Civil Servants: The Whitehall 11 Study.” The Lancet 337. 8754 (1991): 1387-1393. Martin, Emily. “The Egg and the Sperm: How Science has Constructed a Romance Based on Stereotypical Male-Female Role.” Signs 16 (1991): 485-501. . “Toward an Anthropology of Immunology: The Body as Nation State”, in Mario Biagioli, The Science Studies Reader. New York: Routledge, 1999: 358-371. McMullin, Ernan. “Values in Science.” PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association 2 (1982): 3-28. Mills, Charles. The Racial Contract. Ithica: Cornell University Press, 1999. Munson, Ronald. “Why Medicine Cannot Be a Science.” Journal of Medicine and Philosophy 6 (1981): 183-208. Nelson, Hilde Lindemann and James Lindemann Nelson. The Patient in the Family: An Ethics of Medicine and Families. New York: Routledge 1995. 232 Nelson, Hilde Lindemann. “Knowledge at the Bedside: A Feminist View of What’s Happening with This Patient.” The Journal of Clinical Ethics 7. 1 (1996) 25. Nelson, Lynn Hankinson and Jack Nelson, eds. Feminism, Science, and the Philosophy of Science. Boston: Kluwer Academic Publishers, 1997. Nelson, Lynn Hankinson. Who Knows: From Quine to a Feminist Empiricism. Philadelphia: Temple University Press, 1990. . “Empiricism Without Dogmas.” In Nelson, L.H. and Nelson, J ., 95-119. . “Epistemological Communities” in Alcoff and Potter, 121-159. Nettleton, Sarah, Ian Watt, Lisa O’Malley and Philip Duffey. “Understanding the Narratives of People Who Live with Medically Unexplained Illness.” Patient Education and Counseling 56. 2 (2005): 205-210. Ong, L. M. L., et al. “Doctor-Patient Communication: A Review of the Literature.” Social Science and Medicine 40 (1995): 903-918. Ortony, Andrew, ed. Metaphor and Thought. Cambridge: Cambridge University Press, 1993. Oxford English Dictionary Online, http://dictionary.oed.com/, accessed September 9, 2005. Purdy, Laura. “Medicalization, Medical Necessity, and Feminist Medicine.” Bioethics, 15 (2001): 248 - 261. Railton, Peter. “A Deductive-Nomological Model of Probabilistic Explanation.” Philosophy of Science 45 (1978): 206-226. . Explaining Explanation. Diss. Princeton University, 1980. . “Probability, Explanation, and Information.” Synthese 48 (1981): 233-356. 233 Rooney, Phyllis. “On Values in Science: Is the Epistemic/Non-Epistemic Distinction Useful?” PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association 1 (1992): 13-22. Rothman, David J. Strangers at the Bedside: A History of How Law and Bioethics Transformed Medical Decision Making. New York: Basic Books, 1991. Ruben, David-Hillel. Explaining Explanation. New York: Routledge, 1990. Rudner, Richard. “The Scientist ‘Qua’ Scientist Makes Value Judgments.” Philosophy of Science 20 (1953): 1-6. Salmon, Wesley C. Scientific Explanation and the Causal Structure of the World. Princeton: Princeton University Press, 1984. . Four Decades of Scientific Explanation. Minneapolis: University of Minnesota Press, 1989. Scarce, Michael. Smearing the Queer: Medical Bias in the Health Care of Gay Men. New York: Haworth Press, 1999. Schaffner, Kenneth F. “Exemplar Reasoning about Biological Models and Diseases: A Relation between the Philosophy of Medicine and Philosophy of Science.” Journal of Medicine and Philosophy 11 (1986): 63-80. . Discovery and Explanation in Biology and Medicine. Chicago: University of Chicago Press, 1993. Scott, Janny. “Life at the Top in America Isn’t Just Better, It’s Longer”, New York Times Online, http://nytimes.com, accessed May 16, 2005. Sontag, Susan. Illness as Metaphor & AIDS and Its Metaphors. New York: Picador, 2001. Susan M. Wolf, ed. Feminism and Bioethics: Beyond Reproduction. New York: Oxford University Press, 1996. 234 Thagard, Paul. “Explaining Disease: Correlations, Causes, and Mechanisms.” Minds and Machines 8 (1998): 61—78. . How Scientists Explain Disease. Princeton, J N: Princeton University Press, 2000. van Fraassen, Bas C. The Scientific Image. New York: Oxford University Press, 1980. Veach, Robert. “Lay Medical Ethics.” Journal of Medicine and Philosophy 10 (1985): l- 10. Walker, Margaret Urban. “Keeping Moral Space Open.” Hastings Center Report 23 (March-April 1993): 33-40. Webb, Mark Owen. “Feminist Epistemology and the Extent of the Social.” Hypatia 10.3 (1995): 85-98. Werner, Anne and Kristi Malterund, “It is Hard Work Behaving as a Credible Patient: Encounters between Women with Chronic Pain and Their Doctors” Social Science and Medicine, 57 (2003): 1409-1419. Wertz, Richard and Dorothy Wertz. Lying-In: A History of Childbirth in America. New Haven: Yale University Press, 1989. Whitbeck, Caroline. “Ethics as Design: Doing Justice to Moral Problems.” Hastings Center Report 26 (1996): 9-16. Woodward, James. Making Things Happen: A Theory of Causal Explanation. New York: Oxford University Press, 2003. 235 AAAAAAAAAAAAAAAAAAAAAAAAAAAA llllllllllll11111111111111115111111