The Logic of Diagnosis: Peirce, Literary Narrative, and the History of Present Illness

Abstract

This essay presents a theoretical construct upon which to base a working — “pragmatic” — definition of the History of Present Illness (HPI). The major thesis of this essay is that analysis of both the logic of hypothesis formation and literary narrative — especially detective stories — facilitates understanding of the diagnostic process. The essay examines three elements necessary to a successful development of a patient's HPI: the logic of hypothesis formation, based upon the work of the philosopher-logician, Charles Sanders Peirce; the organization of knowledge in relation to structures of narrative; and the feedback necessary to the successful physician-interviewer. It concludes with a systematic description of the design of hypothesis formation within diagnoses.

Keywords:

• abjection
• detective fiction
• medical diagnosis
• narrative logic

I. INTRODUCTION

Without question, the most frequent procedure a doctor performs is a patient interview. Interviews occur over 200,000 times in the professional lifetime of a physician. This verbal and nonverbal interaction forms the backbone of the patient-doctor relationship. In spite of this being the most common procedure, the physician's performance in this arena results in the most common complaints about doctors — “my doctor does not listen to me,” and “my doctor does not seem to care.” Interestingly, patient dissatisfaction almost never centers on the idea of “my doctor doesn't seem to know enough medicine.” It is also common wisdom in medical education that the most powerful diagnostic information is the patient interview, specifically the History of Present Illness (HPI). The HPI is the portion of the patient's story of illness in which the patient tells the symptoms, time frame, relationship of the various symptoms, and his or her interpretation of them. The physician's task in this portion of the interview is to understand the patient's story in the context of the nonverbal clues for such things as anxiety, anger, or depression that are also present. The physician must facilitate a story from a patient experiencing one of these uncomfortable emotions. This cumulative task requires enormous skill, which is not encouraged in much of the curricula of medical school. Many studies in the medical education literature report that medical students become worse at these tasks as medical school progresses. If the doctor can successfully obtain the story (HPI) in a rational sequence, and put it together appropriately, he or she must then apply a rigorous logic to this information in order to arrive at the correct diagnosis. Without a sense of this logic, problems leading to improper diagnoses can — and often do — attend the HPI.

We believe that the process that is described above can be understood — and incorporated into the practices of working physicians — by discerning the logic of diagnosis in relation to literary narrative. There are elements and structures in literary narratives found in novels and short stories that parallel in many ways the narratives that patients tell their doctors (see Vannatta, Schleifer, & Crow, 2005; Charon, Connelly, & Hunsaker Hawkins, 1995; Charon, 2004; Hunter, 1999, esp. pp. 305–308). It is the purpose of this essay to examine the basic methods and skills that allow successful physicians to develop and utilize the HPI to the fullest extent. We examine the logic of hypothesis formation, the uses of a knowledge base, and the reflective feedback of the physician-interviewer in relation to literary narrative. We are assuming that the creation of hypotheses is not simply a process of good “guessing,” but rather is a skill that is acquired. This skill, which Aristotle described as phronesis, is closely related to the kinds of understanding that narrative discourse facilitates. ¹ People — even young children — can distinguish between well-formed and ill-formed narratives with the same skill they distinguish between grammatical and ungrammatical sentences. These seeming intuitions are skills that are learned and can be encouraged, refined, and sharpened. By discovering in what often seem to be lucky, intuitive guesses the “retrospective comprehension” of narrative, physicians can discern the logic of hypothesis formation and develop forms of attention that can make the facilitation of the HPI and subsequent diagnoses less haphazard. An important place for such discovery, this essay argues, is the analysis of detective stories and their relation to Peirce's abduction. Detective stories provide powerful narrative analogues to both histories of present illness and the diagnoses they lead to.

II THE ORGANIZATION OF DIAGNOSIS

The diagnostic process, like the process used by both Sherlock Holmes and Auguste Dupin in the detective stories we examine, has been referred to as “hypothetico-deductive” reasoning (Kaplan, 1995). Diagnosis has at least three elements. The most obvious — and often the seeming only element — is a large knowledge base, empirical knowledge about illnesses and their symptoms. But diagnosis also requires a “method” of hypothesis formation, a procedure for beginning diagnosis. The knowledge base is always brought to bear on the presentation of illness after some method of procedure, even if its method is unconsidered, unconscious, or habitual, and here we will begin — and spend the most time — describing the method or what we are calling the logic of hypothesis formation. The third element of diagnosis is reflection on the presentation in relation to the “hypothetical” possibilities of diagnosis that arise out of the conjunction of the logic of analysis and medical knowledge the physician — like the detective of narrative fiction — brings to patient's story, the HPI. We have characterized this as the “pause and reflection” that stimulates interpretation both in encountering patients and reading narrative (see Vannatta et al., 2005), ² and here we describe it as the third element of diagnosis.

A The Logic of Hypothesis Formation

Abduction

The initial element of diagnosis is the logic of hypothesis formation. As we mentioned, this can be more or less conscious, and most experienced physicians through years of practice have worked out this “logic” so it is virtually habitual. (This is the “practical reason” of phronesis.) At the end of the nineteenth century, the American philosopher Charles Sanders Peirce studied and attempted to formulate the reasoning of hypothesis formation in what he called the logic of “abduction.” Abduction seeks an explanation of a particular fact by finding some salient features of the particular that allow it to be explained by some more general causal principle: “abduction,” Peirce writes, “is the process of forming explanatory hypotheses” (1931–1935, 1958, 5.127). That is, in the logic Peirce formulated, abduction forges a connection between a “case” and a “rule.” Nevertheless, the crucial problem is determining what is “salient.” Moreover, as Harry Frankfurt noted, “Peirce gives no systematically coherent account of abduction” (1958, p. 593). Even so, Peirce does formulate a description of abduction in a syllogism that sounds remarkably like a narrative.

The surprising fact, C, is observed

But if A were true, C would be a matter of course

Hence, there is reason to suspect that A is true. (Peirce, 1931–1935, 1958, 5.189)

That C follows A as a “matter of course” is what we are contending is the narrative element of abduction and, indeed, the reason that self-conscious understanding of narrative can contribute to successful diagnoses. In fact, as we argue, the connection between “case” and “rule” necessarily is apprehended instantaneously.

In what has come to be called his “Bean Bag analogy,” Peirce describes how abduction differs from induction and deduction. Here we are modifying Peirce's syllogism in order to more clearly elucidate its applicability to medicine (and also to reflect the simultaneity of the apprehensions of “case” and “rule”).

Deduction

Rule — All the children in this class have measles.

Case — These children are from this class.

Result — These children have measles.

Induction

Case — These children are from this class.

Result — These children have measles.

Rule — All the children from this class have measles.

Abduction

Result — These children have measles.

Rule — All the children in this class have measles.

Case — These children are from this class. ³

Peirce has suggested that these three kinds of reasoning are all “based upon the idea of an hypothesis” (Burks, 1946, p. 308). “Abduction invents or proposes hypotheses,” Arthur Burks has argued:

it is the initial proposal of an hypothesis because it accounts for the facts. Deduction explicates hypotheses, deducing from them necessary consequences by means of which they may be tested. Induction tests or establishes hypotheses…. “Abduction is the process of forming an explanatory hypothesis. It is the only logical operation which introduces any new idea; for induction does nothing but determine a value, and deduction merely evolves the necessary consequences of pure hypothesis” (5.171). (Burkes quoting Peirce, 1946, p. 303)

In our measles version of Peirce's Bean Bag analogy, the implicit hypothesis in the abduction syllogism is that there is some (logical?) connection between the facts of having measles and being a member of a particular class: the particular implicit hypothesis is that measles is contagious so that it issues from the fact that “these children are from this class.” Abduction invents the hypothesis by suggesting a temporal/causal sequence, which is inexplicit in the Bean Bag analogy: measles, unlike Peirce's white beans, is a condition that has a discoverable history. (As in narrative, it is the causal element that transforms the temporally presented facts into an “instantaneous” apprehension of a “meaningful whole.”) Induction tests this hypothesis: given the (hypothetically) contagious nature of measles it assumes that children from the class have it and examines a reasonable number to confirm its assumption. Finally, deduction explicates this state of affairs. That is, the Rule, “all the children in the class have measles” can be rewritten as an explicit statement of the implicit hypothesis of the abduction syllogism, “in this class, measles is contagious,” leading to a more explicit syllogism:

Deduction

Rule — In this class, measles is contagious.

Case — These children are from this class.

Result — These children have measles.

Abduction transforms the general rule from universal statement to historical event. ⁴

In important ways, then, abduction particularizes the general rule. The aim of abduction is not the logical definition of a particular instance (Deduction) or the articulation of a general rule (Induction). Rather, its aim is to define the relationship between instance and rule, the discovery that a “fact” is a “case,” and for this reason abduction calls for the rewriting of the assumptions brought to a situation in order to situate those assumptions in historical time (“in this class”). As Peirce himself describes it: deduction “is merely the application of general rules to particular cases” (1992, p. 132) and induction “is the inference of the rule from the case and result” (1992, p. 133), while abduction is concerned with the manner in which the rule manifests itself as a cause in a temporal sequence. Thus he argues that abduction “is where we find some very curious circumstance, which would be explained by the supposition that it was a case of a certain general rule, and thereupon adopt that supposition” (1992, p. 135).

Abduction and Narrative

As this suggests, the logic of abduction most closely describes what physicians do when they apprehend the “the surprising fact” or “very curious circumstance” of symptoms as presenting a particular case of an illness within an HPI. If in induction, as Peirce argues, “we conclude that facts, similar to observed facts, are true in cases not examined,” then “by hypothesis [abduction], we conclude the existence of a fact quite different from anything observed” (1992, p. 143). ⁵ Induction, he concludes, “is reasoning from particulars to the general law; [abduction], from effect to cause. The former classifies, the latter explains” (1992, p. 143). The logic of abduction connects empirical Result and the quite different Case. Case is different from the Result because the Case cannot be observed but only hypothesized (these children are not in that class when they present themselves with measles). That is, abduction, faced with the “very curious circumstances” that occasions its reasoning altogether, offers the hypothesis that the event (of “these children”) constitutes a Case. Because abduction aims at discovering a “Case” — discovering, that is, that a particular fact or circumstance is a “case” of a more general rule — it is bound up in the temporalities of narrative and what we have called “narrative knowledge” (Vannatta et al., 2005): the existence of an event as a “case” of a more general proposition is analogous to the existence of an event as an “episode” in a narrative that presents itself as a “meaningful whole” (Vannatta et al., 2005; see also Hunter, 1999, pp. 306–308).

It is the assumption of this essay — as it is of our DVD publication Medicine and Humanistic Understanding — that narrative is defined by the fact that it presents a “meaningful whole,” that one can grasp a global sense of a narrative in the same way we grasp the meaning of a sentence that is not reducible to any one element of the sentence. In Medicine and Humanistic Understanding we argue that the explicit narratives of stories and the implicit narratives in some poems transform simple, seemingly isolated phenomena into meaning; they create what A. J. Greimas describes as “the still very vague, yet necessary concept of the meaningful whole set forth by a message” (1983, p. 59). Such a “meaningful whole” is the overall sense or point of a story — its “thought,” its “aboutness” — the meaning we take away from it, the “moral” of the tale, and/or even its sense of overall “genre.” Thus, we say Hamlet, Oedipus, and even the death of John Kennedy are “tragedies” despite the fact that they are so different. We make this judgment (or simply have a more or less unarticulated sense of their similarity) because each of these very different narratives of Hamlet, Oedipus, and Kennedy configure or “grasp” a series of events — intellectual promise, prominent political power, unforeseen yet recognizable violence — in a manner that conveys or provokes particular cognitive and emotional responses. In this way, a “vague” sense of a whole unified meaning emerges from the elements of narrative presented. Similarly, we say that the combination of red dots on the skin, persistent fever, nasal discharge, and evidence of contact with others possessing these attributes might well be measles (see Vannatta et al., 2005, Chapter 2, screens 12–13 for a version of this passage). A sense of the meaningful whole of narrative is particularly prominent in detective fiction, whose explicit goal is to present a fact or a situation that allows the many events of the narrative to fall into place — the be explained — as elements within a larger, “whole” pattern of action, of cause and effect. ⁶

For this reason, the structure of abductive reasoning is described by Auguste Dupin in “Murders in the Rue Morgue”: “There is no evidence of the killer's escape; the killer is not on the premises: therefore, there must be evidence of the killer's egress” (Poe, 1985, p. 254). That is, in abduction the “surprising fact” (no evidence of the killer's escape) must be accounted for as a “case” that follows from a hypothesized rule in relation to the facts (“result”) — accounted for, that is, as a “matter of course” in a narrative. Part of the procedure of taking a patient's history is to rule out things that haven't been observed and discover “pertinent negatives.” Physicians ask patients what they haven't noticed: “did that tenderness on your arm occur after a fall?” for instance; or “how long after your tooth began to hurt did this low, persistent fever begin?” It is the inclusion of such “negative” evidence — that is not “observable” in the manner of positive facts — that in part makes the explanation of cause different in kind from logical reasoning and discovering facts (though both logic and discovery can contribute to such explanations).

Such explanation is closely connected to the working of narrative understanding, as opposed to the logical understanding of deduction and the empirical understanding of induction. ⁷ Neither the logical conclusion that these children have measles nor the conjecture that all the children in the class have measles tells a story in the way that the logic of abduction tells a story in its connections of effects and causes, that “these children have measles because they are from this class.” Narrative organizes disparate events into complex action in which an array of happenings becomes the sequence of action that can be apprehended — “retrospectively” and “instantaneously” — as a whole. In its most basic form, narrative presents an initial situation, a change or reversal of that situation, and — most importantly — retrospective comprehension of what has taken place. Thus, as J. Hillis Miller notes, “the minimal personages necessary for a narrative are three: a protagonist, an antagonist, and a witness who learns” (1995, p. 75), though he adds that the protagonist, antagonist, or reader may be the witness. Narrative comprehension accomplishes what has been called the “synthesis of the heterogeneous” by organizing a series of events “into an intelligible whole, of a sort such that we can always ask what is the ‘thought’ of this story” (Ricoeur, 1984, pp. ix, 65): as we have noted, it grasps a meaningful whole. That is, for narrative the whole is more than the sum of its parts; more than simply a collection of “data” added together. Umberto Eco describes this process in terms of Peircean semiotics as the transformation of “a disconnected series” into “a coherent [textual] sequence” that allows us to recognize the “‘aboutness’ of the text which establishes a coherent relationship between different and still disconnected textual data” (1983, p. 213). And Peirce himself argues that “the essence of an induction is that it infers from one set of facts another set of similar facts, whereas hypothesis [or abduction] infers from facts of one kind to facts of another” (1992, p. 150).

The second order of fact pursued by the logic of abduction, Peirce suggests, is “very frequently a fact not capable of direct observation” (1992, p. 150). Similarly, a “medical fact” not capable of direct observation is the “cause” — the illness — underlying symptoms that patients present to physicians in the narratives of present illness they articulate. It is on this level of analysis that narrative understanding allows the grasping of the connection between instance and rule, between effect and cause. The rule articulates the “aboutness” Eco describes; it articulates a meaning — in the case of medicine, a diagnosis of a “general” disease or condition, or in the case of measles the “contagion” implicit in the “Rule — All the children in this class have measles” — to which the particular symptoms (instances) are related as a Case. This can only be the hypothesis of abductive logic because although the general Rule is given (as in deduction), its implications are not “unpacked”; and although the Case is also given (as in induction), like the Rule, the Case has further, implicit significance. That is, it is precisely the relationship between Rule and Case that must be established from the “given” data of possible rule and particular instance, and that relationship exists as an implicit narrative: “these children have measles because they are from this class.”

Peirce suggests the narrative nature of abduction when he distinguishes abduction from induction in terms of the fact that induction begins with “objects” and “facts” while abduction begins with the “characters” or “characteristics” of phenomena (for instance, the attribute of measles or, better, the attributes small red dots on the skin, fever, nasal discharge) which have to be configured within particular categories or frameworks. “Hypothesis [abduction],” Peirce writes, “has been called an induction of characters…. [and] characters are not susceptible of simple enumeration like objects; [rather], characters run in categories” (1992, p. 140). It is the work of abduction, Peirce goes on to argue, to isolate “a single line of characters, or perhaps two or three, and … take no specimen at all of others” (1992, p. 140). That is, the work of abduction is not simply to apprehend an instance as a Case, but to apprehend simultaneously the Case and the Rule by discovering the framework in which the relationship between event and understanding can be defined as that of a Case and a Rule, that a chosen fact is salient (i.e., a “case”). Paul Ricoeur has described this particular mode of understanding as “configurational”: “the configurational mode,” he writes, “puts its elements into a single, concrete complex of relations. It is the type of comprehension that characterizes the narrative operation” (1984, p. 159). Umberto Eco describes this operation more concretely: “The real problem is … how to figure out both the Rule and the Case at the same time.” In the Bean Bag analogy, he continues, “Peirce could have decided that the crucial element was not where [which bag] those beans came from, but — let us say — who brought them there” (1983, p. 203). The “crucial element” is the framework of understanding that allows a fact or event to be configured and comprehended as a Case. The “thought” or genre of a story is such a framework; its comprehension is the work and the goal of narrative. In medicine such comprehension is the diagnosis, the “explanation” of the symptom, that suggests both cause and treatment. The most important occasion for this explanation — for those trained to listen in certain ways and pursue what Sherlock Holmes calls “reasoning backwards” (Sebeok & Umiker-Sebeok, 1983, p. 39, citing “A Study in Scarlet”) — is the History of Present Illness.

B The Power of Knowledge

We have been discussing the initial element of diagnosis, the logic of hypothesis formation, since our aim is to make this — often unconscious or unconsidered — aspect of diagnosis more discernable and thus to allow for its more self-conscious pursuit. But the most obvious element of diagnosis, as we mentioned, is the knowledge base. Knowledge is important because it is necessary to validate old hypotheses and to generate new ones by suggesting the possible frameworks of understanding we mentioned above that infuse meaning into otherwise trivial signs. In the terms we have been developing, it is the knowledge base that often allows the particular instance to be apprehended as a Case of a more general situation. The functioning of such knowledge is clear in one of the many little narratives Peirce presents in “Deduction, Induction, and Hypothesis,” a narrative that reads very much like the stories of Arthur Conan Doyle, Peirce's contemporary.

A certain anonymous writing is upon a torn piece of paper. It is suspected that the author is a certain person. His desk, to which only he has had access, is searched, and in it is found a piece of paper, the torn edge of which exactly fits, in all its irregularities, that of the paper in question. It is a fair hypothetic inference that the suspected man was actually the author. The ground of this inference evidently is that two torn pieces of paper are extremely unlikely to fit together by accident. (1992, pp. 139–140)

Later in this paragraph Peirce argues that “if the hypothesis were nothing but an induction, all that we should be justified in concluding … would be that the two pieces of paper which matched in such irregularities as have been examined would be found to match in other, say slighter, irregularities. The inference from the shape of the paper to its ownership is precisely what distinguishes hypothesis [abduction] from induction, and makes it a bolder and more perilous step” (1992, p. 140). In this example, the key difference between induction and abduction is conditioned by the knowledge of another “kind” of fact, namely the knowledge that only the suspect “has had access” to the desk. This fact — which Peirce only mentions here in passing and does not mention again in his argument — is different in kind from the evidence of the torn paper both because it is focused on a different object from the inductive conclusions about the paper and also because it is not observable: like the prior existence of Napoleon that Peirce also presents as simply a hypothesis (1992, p. 135), it is a “fact” that is not capable of being empirically observed. Neither are the abstract, “general” illnesses observable except as an inference based upon a particular combination of symptoms. Such knowledge seemingly “outside” the parameters of the logical problem allows the bolder, more perilous activity of abductive logic.

It is the job of medical training to give physicians possession of such “second-order” knowledge. More dramatically, both Edgar Allan Poe and Arthur Conan Doyle emphasize the great breadth and depth of knowledge their detectives possess in the stories we examine. Signs at the murder scene in Poe's “Murders in the Rue Morgue” such as the position of the body in the chimney, the strange marks around the neck of one of the victims, and the ribbon found by the lightning rod, were meaningless to the police, but not to Dupin. This evidence would be trivial details if not for Dupin's knowledge of wild animals, human anatomy, the navy and its sailors. Instead, they are crucial in Dupin's understanding that the deaths in the story are irrational violence rather than murder. Similarly, Holmes's knowledge of tobacco and cigars in “The Resident Patient” is crucial to his apprehension that the death in that story is murder rather than suicide because it suggests that there were other people involved. In both of these instances, the global category of “murder” (vs. “irrational violence” or vs. “suicide”) defines the “aboutness” of the stories. In these instances, “violence” and “murder” are what Peirce calls the “characters” (or “characteristics”) which abduction focuses upon. Such a focus transforms classification (i.e., the empirical fact of “death”) into explanation (i.e., the understanding of that death as “irrational violence” in Poe and “murder” in Doyle).

C Reflection

The third element of diagnosis is reflection. The physician shares with the detective the need to attend to the manner of their own interpretive action as well as the condition of the patient. The process of playing one's hunch or hypothesizing a connection between a particular instance and a particular rule must be followed by testing the validity of that connection in relation to some overall “meaning” — the “aboutness” of a narrative, the particular “illness” behind symptoms — in which the instance is seen to be a Case (of a disease or of a crime such as “murder”). This is the ability of the detective, physician, or abductive logician to recognize the “aboutness” of the facts — the ability to apprehend them as evidence — in order, as Eco says, to establish “a coherent relationship between different and still disconnected … data” (1983, p. 213). We are calling this “reflection” because the hypothesis-generation of abduction doesn't follow logical rules that could be programmed into a computer (to play chess, for instance); nor does it follow empirical rules of evidence collection and classification, that could also be programmed, logarithmically, into a computer (see Hunter, 1999, pp. 308–309). Rather, abduction follows feedback rules such as those articulated by information theory — and developed in the experience and skillful practices of reading narratives — in which skills grow through experience, which cannot be programmed but only accumulated (though John Holland [1995] suggests even this accumulation and “characterization” of experience can be programmed). The detective who completes the loop of hypothesis generation and hypothesis testing is the detective who learns and improves. Likewise, the diagnostician who is reflective in his or her approach — who makes the “facts” of her or his activity part of the “logic” of diagnosis — will continue to improve with experience. Such reflection calls upon “characterization.” That is, all observations are not necessarily interchangeable; they can be of different modal orders, particularly the modalities of seeing and hearing, both of which should be resources of the HPI. The young physician has a great knowledge base with little experience to test it. The experienced physician, who has not been reflective of process and outcome, has failed to learn from his experience the kinds of phenomena to attend to. Like the police in detective stories, such “experienced” physicians too easily fail to apprehend details that disrupt preconceived generalizations, in part because they think all (empirical) observations are the same.

D Abduction and Diagnosis

These three elements of diagnosis — hypothesis formation, testing against a knowledge base, and reflective understanding of the process — nicely fit into Peirce's schema of logical inferences of invention, testing, and explication, which we examined earlier. But equally important, they also describe what we do when we read and what physicians do when they encounter the patient's HPI. Thus in an interview Dr. Rita Charon has noted that:

stories have plots. Illnesses have symptoms. In a strange way, when a doctor is trying to diagnose a patient, when a doctor is hearing about many symptoms, events, sensations, feelings, things out of the ordinary from a patient describing new symptoms, in a funny way what the doctor does in diagnosis is pretty much what he or she does in reading for the plot …. Now, … whether it's a joke somebody tells me or a long, complex novel like Beloved, the activity of the reader is to register each event, whether or not they're told chronologically, and to reconfigure them using our imagination and our memory so that they make at least provisional sense.

And, so, isn't that what we do, again, as I'm sitting in the office listening to the woman with abdominal pain …. [A]nd I, as the diagnosing doctor, have to somehow register these events, configure them in my mind so that they make provisional sense. (quoted in Vannatta et al., 2005, Chapter 2, screen 52)

Charon is describing the ways in which readers and doctors both make hypotheses about what they hear and encounter, gather new evidence, and reflectively reconfigure understanding, again and again. Detective fiction is about these encounters, as Auguste Dupin or Sherlock Holmes encounter stories, create hypotheses, test them, reflect upon then, over and over again. But this happens whenever we encounter narratives.

Physicians encounter narratives. Thus, in one incident, a resident admitted a middle-aged woman from Wewoka, Oklahoma, during an extraordinarily busy night. As he entered her room, the woman was buried in covers, her face without expression, skin sallow in appearance as if she were chronically ill or depressed. There were no family members — they complained her thinking was “messed up” and had left at midnight — the room was barren except for a small pile of mostly worn-out clothes, a pair of rayon stretch pants pulled through the rounded handles of a naugahyde purse sitting on a chair. Her responses were short, usually not to the point, and irritated. Feeling angry at her failure to help in the diagnosis, the resident hurried through her narrative of recent events, past history, and systems review. When asked, she specifically denied taking any medication.

Still, it was clear to the resident that the patient's problem was that of hyponatremia, the dilution of the sodium concentration in the blood. Since the blood is basically a salt water solution, reasons for the decreased sodium concentration should not be an unsolvable riddle: medications, and their affects on brain hormones or kidney, an under-active thyroid gland, psychogenic water drinking, adrenal insufficiency, congestive heart failure, chronic renal failure, low albumin levels in the blood, ascites (free water in the abdominal cavity), stress, pain, vomiting, diarrhea, the syndrome of inappropriate antidiuretic hormone, each was a possible “cause” of her condition, a Rule that would allow the woman's condition (Result) to be apprehended as a Case. Yet, neither the resident's questions nor his physical examination provided an answer to the cause — and subsequent treatment — of her sodium concentration.

When he met with the attending physician the next morning, all the resident could report was that the patient was a “bad historian.” “Was she a bad historian,” the attending physician asked, “or were you an inadequate interviewer?” The attending physician questioned the patient again, and again asked if she were taking medication and whether she had something to add to her history. “Why do you keep asking me all these questions?” she asked as she glanced at the nightstand, then down toward the end of the bed. Afterwards, the attending physician mysteriously announced to the resident and interns that in her purse she had chlorthalidone, a diuretic that in this age group commonly causes hyponatremia. When they returned to the patient, she was curled up under the bedclothes with the blanket over her head, and the purse that was evident the night before was nowhere in sight. The attending physician asked the patient to get her purse “so I can look at the pills in it.” She rolled over toward the bedside table, pointed to it and told the intern she could get it out for her. The purse contained, among other things, four pill bottles, a thyroid medication, a tranquillizer, a pain medication, and the diuretic, chlorthalidone. “Do you take these pills every day?” “Most days. I didn't take any yesterday because I felt bad.” She lay back down, pulled the covers over her head, and said she wanted to be alone.

A narrative such as this — several like it are presented as vignettes in Medicine and Humanistic Understanding (see “vignettes” in the index) — demonstrates the ways that physicians, in their encounters with patient narratives, pursue hypothesis formation, test the hypothesis against a knowledge base, and develop a reflective understanding of the process. The resident possesses a knowledge base, yet lacks the experience to interpret the patient's nonverbal language of looking at the nightstand. Moreover, while the resident has a knowledge base, it is not quite as experienced and deep as the attending's knowledge base: he does not know how commonly hyponatremia is the result of chlorthalidone. In fact, he has not yet gained enough experience — enough practical reason — upon which to develop as keen a hypothesis as the attending physician. ⁸

III THE DETECTIVE STORIES

We are suggesting that the hypothesis-generating process of diagnosis — a process that informs the HPI — is closely connected to the manner in which that narrative discourse creates meaning or significance. For this reason, we believe, physicians can learn significant skills — they can, in fact, acquire the reflective “experience” we have described in relation to diagnostics — by studying narratives (see Hunter, 1991; Hunter, 1999; Charon et al., 1995; Charon, 2004). As we have mentioned — and as the narrative of the woman with hyponatremia suggests — the narrative genre that most explicitly focuses upon abduction and hypothesis formation is that of the detective story or novel. In a detective story, the detective is a witness who learns from the narratives he or she encounters: the detective inhabits a world of rules, a system of general laws, and is presented with an array of particular narrative and evidentiary instances which need to be apprehended as connected to a particular rule or a particular characterization of events. Similarly, a doctor is a witness who learns from the narratives he or she encounters: inhabiting a world of rules, a system of general classes and subclasses of diseases, the physician is presented with a particular narrative (explicit and implied) and physical evidence (a symptom or set of symptoms) which need to be apprehended as connected to a particular rule, to be seen as a Case of a particular disease or condition. Thus it is no accident that Peirce's exposition of abduction repeatedly offers small narrative situations in order to present the distinction he is making between inductive and abductive logic (both of which he opposes to deductive logic).

The classic detective story is considered to have been developed by Edgar Allan Poe and Arthur Conan Doyle, the latter of whom was, in fact, a trained physician. The two detectives they created — Poe's Dupin, created in the mid-nineteenth century, and Doyle's Sherlock Holmes, created nearly a half century later — are model detectives who “witness” stories and learn. In “The Murders in the Rue Morgue,” Poe presents the detective to be a man of unusual talents. He possesses enormous knowledge and is also a master of understanding human emotions by studying facial expressions and other body language: this, indeed, demonstrates his ability to apprehend coherent relationships among different and disconnected facts. (“Most men,” Dupin [Poe, 1985, p. 249] says, “wore windows in their bosoms.”) Moreover, the detective is able to use logic in a manner uncharacteristic of most people. Early in this story, Dupin demonstrates his remarkable ability to hypothesize by guessing that his companion, the narrator, was thinking of a bad tragic actor named Chantilly. “Tell me, for Heaven's sake,” the narrator exclaimed, “the method — if method there is — by which you have been enabled to fathom my soul in this matter” (Poe, 1985, p. 250). Dupin goes on to explain the “links” of the narrator's thought: “the larger links of the chain run thus — Chantilly, Orion, Dr. Nichols, Epicurus, Stereotomy, the street stones, the fruiterer” (Poe, 1985, p. 250). Reasoning backwards — reasoning reflectively — Dupin “remembers” what the narrator forgets so that he is able to create the larger links of narrative where the narrator had only experienced disparate “facts.” The narrator's collision with the fruitier, which began his chain of thought, is a different order of fact from the narrator's final thought about the actor; what Dupin does is demonstrate that different and disconnected data can be apprehended together in relation to a single train of thought. This is the “aboutness” of the narrative; it allows Dupin to apprehend disparate facts as “a single line of characters” and transform events into explanation.

That is, Dupin makes every event into a “Case” or signifying instance of his companion's character. This passage goes on to meditate on this train of thought by presenting a narrative where, before, there was a set of facts that presented such an “apparently illimitable distance and incoherence between the starting point and the goal” (Poe, 1985, p. 250) that the narrator did not even remember all of them. Dupin makes these facts into a story. “We had been talking of horses,” Dupin explains,

“… if I remember aright. Just before leaving the Rue C—. This was the last subject we discussed. As we crossed into this street, a fruiterer, with a large basket upon his head, brushing quickly past us, thrust you upon a pile of paving-stones collected at a spot where the causeway is undergoing repair. You stepped upon one of the loose fragments …. I was not particularly attentive to what you did: but observation had become with me, of late, a species of necessity.

“You kept your eyes upon the ground [until we encountered] … riveted blocks. Here your countenance brightened up, and perceiving your lips move, I could not doubt that you murmured the stereotomy, a term very affectedly applied to this species of pavement. I knew that you could not say to yourself “Stereotomy” without being brought to think of atomies, and thus of the theories of Epicurus; and since, when we discussed this subject not very long ago, I mentioned to you how singularly, yet with how little notice, the vague guesses of that noble Greek had met with confirmation in the late nebular cosmogony, I felt that you could not avoid casting your eyes upward to the great nebula in Orion …. (Poe, 1985, pp. 250–251)

Dupin goes on to note that a negative review of the actor Chantilly the previous day cited a Latin line that also referred to Orion, that the narrator knew of this reference, so that “it was clear, therefore, that you would not fail to combine the two ideas of Orion and Chantilly” (Poe, 1985, p. 251).

In this passage Dupin emphasizes the extraordinary powers of observation of human behavior needed by the detective and in so doing he confuses, as Peirce notes, the empirical classifications of induction and the meaningful explanations of abduction. Thus even while he asserts his power of observation, Dupin describes the process of “guessing,” or hypothesis formation, that he uses. Dupin makes hypotheses and then, “reflectively,” tests them against experience — which, just as for the physician, is someone else's experience. Unlike the police and physicians who do not diagnose crimes and illnesses particularly well, Dupin does not commit himself to his guess prematurely. Rather, his observations are both educated and open to inclusion of different “lines” of character. Peirce calls the method of abduction a “fair guess” (1992, p. 134), but important to our discussion of the process of abduction in relation to both the detective's method and the method of the diagnostician is the ways in which its skills are repeatable and teachable. Thus, in the hyponatremia narrative, the attending physician is using and teaching through example the “abductive” method of Dupin. ⁹

It is clear from the ways in which Arthur Conan Doyle repeats the detective techniques that Poe introduced that the process of abduction is repeatable. At the beginning of “The Resident Patient,” as at the beginning of “Murders in the Rue Morgue,” Holmes, watching Dr. Watson, guesses at the narrator's thoughts and explains his guess as a form of deduction. This process is not really deduction but hypothesis generation (abduction) just as Dupin's process is more than simply the empirical observations of induction. (This is made clear by the fact that Dupin's observations are aural rather than visual: he apprehends and remembers words rather than images.) Holmes's logic is not deduction in the strictest sense because there are other data (lines of thoughts) which might have been present and which he chooses to ignore in presenting his explanation. Moreover, his guess turns out to be based on a good hunch and works because Holmes knows Watson so well. This knowledge is a different “kind” of fact. Holmes's many conversations with Watson in the recent past allow him to gather together a large amount of information that might otherwise seem simply different and disconnected facts. That is, Holmes combines knowledge and personal experience to articulate a logic of abduction. The fact that the logical process used by the detectives in these stories and by the physician are similar allows us to think that the “guesses” of abduction are educated guesses — Peirce's “fair” guesses — and suggests that detective stories are especially effective in the education of the diagnostician.

In a passage from “The Resident Patient” (which Doyle also uses, almost verbatim, in “The Cardboard Box”) Holmes presents the kind of reasoning Dupin exhibits in narrating Watson's chain of thought (he even explicitly mentions Poe's story). “After throwing down your paper,” Holmes says,

“… you sat for half a minute with a vacant expression. Then your eyes fixed themselves upon your newly framed picture of General Gordon, and I saw by the alteration in your face that a train of thought had been started. However, it did not lead very far. Your eyes turned across to the unframed portrait of Henry Ward Beecher, which stands upon the top of your books. You then glanced up at the wall, and of course, your meaning was obvious. You were thinking that if the portrait were framed it would just cover that bare space and correspond with Gordon's picture over there.”

“You have followed me wonderfully.” (Doyle, 1986, p. 580)

Holmes goes on to discuss more of Watson's thoughts based on the interpretation of body language, knowledge of Watson's old war wounds, and other knowledge he possesses about his friend. Here again, Holmes emphasizes his powers of observation without analyzing the ways in which observations of different orders of “fact” are brought together to generate an explanation, what we have variously called a “line of thought” or the “aboutness” of a collection of facts. In this narrative Holmes, like Dupin or an experienced attending physician, is “reasoning backwards”: he is making sense of phenomena by supplying their “causes” that relate them to one another; he is connecting Case and Rule. In these instances, the detectives use abduction to “guess” what another person is thinking, even if Holmes's observations are unrelentingly visual (note his attention to Watson's eyes) even as Dupin's are aural (attending to the narrator's words). In any case, such abductive “guesses,” of course, should happen in the practice of medicine. In medical school, we teach courses in human behavior to inculcate in medical students the ability to read body language and emotional states as the detectives do in these stories. These observations and their analysis help the clinician develop hypotheses about the patient's motives, needs, and desires.

In both of these archetypal detective stories these preambles are presented to frame the manner in which each detective observes details of the death scene and of the circumstances surrounding the deaths. In both instances, the detective solves the “murder” when the police have failed by erroneously jumping to the obvious conclusion too early, before all the data are analyzed, classified, and, most of all, characterized carefully. That is, the police follow methods of induction which leads them to exclude information as irrelevant because they have, from the start, a particular theory in view. In this, they aim at classification rather than explanation. In the case of the “Rue Morgue” they assume “murder” is the controlling framework of understanding rather than irrational violence, and thus they cannot see (or really hear) what is there to observe, that the non-language of the supposed colloquy between killer and accomplice suggests the killing was perpetrated by an orangutan rather than a person. Dupin says that they have fallen into the gross but common error of confounding the unusual with the abstruse. Similarly, in “The Resident Patient” they assume suicide instead of what Holmes calls “a deeply planned and cold-blooded murder” (Doyle, 1986, p. 594) because they cannot see what there is to observe. (In the case of the woman with hyponatremia, the resident assumes the patient is withholding evidence and leaves it at that.) The kind of mistake the police make — the assumption that the goal of diagnosis is classification rather than explanation — is that of clinicians who narrow the differential diagnosis too early, failing to account for all the data, both positive and negative, and failing to attend to the manner or modality of their own observations. When Watson says to Holmes that what the detective saw “was quite invisible to me,” Holmes replies: “Not invisible, but unnoticed, Watson. You did not know where to look, and so you missed all that was important.” Holmes goes on to tutor Watson: “Never trust to general impressions, my boy,” he says, “but concentrate yourself upon details” (Sebeok & Umiker-Sebeok, 1983, p. 21, citing “A Case of Identity”). That is, narrative comprehension teaches physicians that diagnosis is a form of explanation, not a method of classification.

IV THE DESIGN OF HYPOTHESIS

What Holmes means by “details” and Dupin by “observation” conditions abduction rather than induction. “Abduction,” Peirce writes,

makes its start from the facts, without, at the outset, having any particular theory in view, though it is motivated by the feeling that a theory is needed to explain the surprising facts. Induction makes its start from a hypothesis which seems to recommend itself, without at the outset having any particular facts in view, though it feels the need of facts to support the theory. Abduction seeks a theory. Induction seeks for facts. In induction the study of the hypothesis suggests the experiments which bring to light the very facts to which the hypothesis had pointed. (1931–1935, 1958, 7.218)

This description of abduction has an excellent parallel in the diagnostic process of medicine. The attending physician, faced with the “fact” of hyponatremia, finds a hypothesis that recommends itself based upon his understanding of the additional fact (not quite “known” by the resident) that chlorthaladion is a commonly prescribed diuretic that commonly causes hyponatremia. Thus, the physician's wider “theory,” like those of Dupin and Holmes, is that patients' stories include nonverbal signs that can and should be read. This theory transforms the “abstruse,” such as the unusual presentation of a common illness, or an uncommon illness manifesting common symptoms, into simply a surprising fact — a different order of fact — that can be tested in the cause-and-effect relationship of narrative: in this instance, that the patient's nonverbal responses might well indicate that what seems mysteriously abstruse might only be “unusual” resistance to the medical establishment (though what the physician teaches the resident and interns is that such resistance is not, in fact, so unusual). Similarly, in “The Resident Patient” the victim was found hanging in his room alone, but Holmes esoteric knowledge of tobacco allows him to hypothesize (abduct) that other people were in the room with him and the death was a different order of fact from that of “suicide.” And in the “Murders in the Rue Morgue,” Dupin's knowledge of naval life (the knot in the ribbon found at the “murder” scene) combined with his sense that the reported “abstruse” languages of the “killers” might be the “unusual” noise of an orangutan allows him to hypothesize that the killings were not “murder.”

Both of the stories — like that of the woman with hyponatremia — provide rich and interesting narrative experience for physicians. They can be used to demonstrate the process of developing the HPI, with its goal, much like that of the detective story itself, to explain seemingly disparate facts. They also emphasize the importance of the psychosocial observations of both Dupin and Holmes — that different order of fact — to suggest the usefulness of attending to the nonverbal cues of patients in relation to the content of their HPI. Finally, they present the logic of hypothesis formation in the activities of the detectives that can be explicitly analyzed in relation to other forms of logic. Thus, the most important aspect of such readings for practicing physicians is the possibility they present for the systematic analysis of the “guesses” of diagnosis. The three-part description of diagnosis we presented at the beginning of this essay — its logic, its relationship to different kinds of knowledge physicians bring to their encounters with patients, and the necessary reflective nature of its understandings — all underline the ways that the narrative HPI can facilitate the explanation of facts. That is, diagnosis has a particular relationship to narrative insofar as narrative presents a retrospective mode of understanding — the “reasoning backwards” of reflective comprehension — that can, in fact, be “experienced” simply by reading. For this reason, the systematic apprehension of the ways that stories create explanations is of the utmost importance to the practice of medicine.

Notes

1. For an important discussion of Aristotle's phronesis, which she translates as “practical reason,” in relation to the usefulness of narrative to medical practices, see Kathryn Montgomery Hunter, “Narrative, Literature, and the Clinical Exercise of Practical Reason.” In this essay, Hunter describes “the circular, hermeneutic procedure that ensues [in phronesis as] … familiar to lawyers, physicians, and moral reasoners, all of whom are required to fit the overarching laws of their disciplinary world view to the particular circumstances — called in each instance a ‘case.’ Neither induction nor deduction, this nevertheless quite rational procedure is what the pragmatist C. S. Peirce termed ‘abduction’” (1999, p. 305). Phronesis is also discussed in relation to clinical practice and narrative in Vannatta, Schleifer, and Crow, Medicine and Humanistic Understanding, especially Chapter 5.

2. Medicine and Humanistic Understanding is an interactive DVD-ROM of more than 700 screens that includes a search engine. Therefore, references to this publication will not have page numbers, but offer — either explicitly or implicitly — key words to search. In this instance, “pause and reflect” will presents screens discussing the ways in which reflection functions in both literary and medical interpretation.

3. The following is the “Bean Bag analogy” as it appears in Peirce's essay “Deduction, Induction, and Hypothesis.” We are following this “translation”: “beans” = “children”; “bag” = “class”; “white” = “measles.” We have also modified “Abduction” to reflect the connection between “case” and “rule” in relation to the “result,” which, in medicine, is the presenting symptom.

• Deduction

• Rule — All the beans in this bag are white.

• Case — These beans are from the bag.

• Result — These beans are white.

• Induction

• Case — These beans are from this bag.

• Result — These beans are white.

• Rule — All the beans from this bag are white.

• Abduction

• Rule — All the beans in this bag are white.

• Result — These beans are white.

• Case — These beans are from the bag. (Peirce, 1992, p. 134)

Since this essay is a thorough early description of “hypothesis” or “abduction,” we are citing an easily accessible source. (The essay also appears in Peirce, 1931–1935, 1958, Volume 2.)

4. Ilkka Niiniluoto (1999, S447) notes that “inference to the best explanation” — a twentieth-century description of abduction (see note 7) — “is always restricted to a set of historically given or formulated hypotheses.”

5. For a thoughtful discussion of accounting for phenomena that are not — and often cannot be — observed, see James Bogan and James Woodward (1988), and especially the distinction they make between “data” and “phenomena” — the distinction between “what is uncontroversially observable (data)” and “what theories explain (phenomena or facts about phenomena)” (1988, p. 314). Instances of phenomena, they note, “can occur in a wide variety of different instances or contexts. This, in turn, is closely connected with the fact that the occurrence of these instances is (or is plausibly thought to be) the result of the interaction of some manageably small number of some causal factors, instances of which can themselves occur in a variety of different kinds of situations …. Data are, as we shall say, idiosyncratic to particular experimental contexts, and typically cannot occur outside of those contexts …. Phenomena, by contrast, are not idiosyncratic to particular experimental contexts. We expect phenomena to have stable, repeatable characteristics which will be detectable by means of a variety of different procedures, which may yield quite different kinds of data” (1988, p. 317). This distinction, implicit in Peirce's conception of abduction — and especially, as we shall see, in his conception of the “character” or “characteristics” of phenomena that abduction focuses upon — might help to clarify the distinction between “symptoms” and “illness” in medicine: symptoms are data; illness is phemonenon.

6. This is why the essays collected in The Sign of the Three, edited by Umberto Eco and Thomas Sebeok, can consistently focus on the relationship between Peirce's semiotics — and especially his sense of abduction — and the narratives of Sherlock Holmes. In a technical philosophical analysis of abduction, Ilkka Niiniluoto also notes that “a historically interesting application of abduction as a heuristic method can be found in classical detective stories” (1999, p. S440). Throughout Doctors' Stories: The Narrative Structure of Medical Knowledge, Kathryn Montgomery Hunter describes the parallel structure of medical diagnosis and detective fiction. “The medical case,” she writes, “the central narrative account of the study and diagnosis of disease in an individual patient, developed along with that most modern of Western literary forms, the detective story” (1991, p. 21). She goes on to say that Sherlock Holmes's “method requires the retrospective construction of a hypothetical narrative in order to work out the relation of the clues to one another within an acceptable chronology. Within the published story written by Arthur Conan Doyle and told by Watson, the action ends with Holmes telling the story he has constructed. Holmes's story is diagnostic, a narrative reconstruction that aims to recapture lost time and unobserved deeds …. The diagnostic skill that is the focus of a physician's education bears striking similarities. Like Sherlock Holmes's narrative reconstruction of the crime, the physician's medical version of a patient's story is the narrative embodiment of a diagnostic hypothesis, the reconstruction of what has gone wrong” (1991, p. 24). See also Montgomery (2000) for a discussion of clinical practices and detective stories.

7. This catalogue is suggested in many post-Peircean studies of abduction and explanation. For instance, in their discussion of the nature of explanation — and especially scientific explanation — in “Saving the Phenomena,” Bogan and Woodward are careful to describe explanation as entailing “detailed causal stories” (1988, p. 324) as well as the “classical deductive-nomological model” of mathematics (1988, p. 324). Their example of a story or narrative explanation is an account of the cause of Parkinson's disease, and they claim that the deductive model often entails “idealizations and approximations” (1988, p. 324n). In an earlier discussion of “inference to the best explanation” (1965) — his description of the logic of abduction (1965, p. 88) — Gilbert Harman argues that “enumerative induction” is a special case of abduction (“inference to the best explanation”).

8. Kathryn Montgomery Hunter describes the parallel between the process of reading and the “practical reason” of the clinician: “like reading itself,” she writes, “the exercise of practical reason is an act of interpretation”; it is a “process intertwined with history, identity, culture, and life-meaning” (1999, p. 307). Describing the “logic” of abduction, Ilkka Niiniluoto also suggests parallels between reading narrative and the differential diagnosis: “For example, suppose we are looking for an explanation of the death of a certain person, and there are available several possible causes, H, [Hacute], … of her death. However, if the case is given a fuller description, including facts about her life and the symptoms in her body, it may be that many of the potential hypotheses fail to give a causal explanation any more. In the limit, it may happen that only one of the potential hypotheses is left, and this is then certainly the ‘best’ explanation in this situation” (1999, p. S444). It is striking that Niiniluoto's example — in a highly technical philosophical paper — is, in effect, the problem of both a detective story anda medical diagnosis.

9. Kathryn Montgomery Hunter has closely studied the relations — and parallels — between the activities of detectives and physicians in Doctors' Stories (1991). Here, we are triangulating these parallels with Peirce's logic. The key question, as Arthur Burks noted many years ago, is this: “can there be a logic of discovery? For are not discovery and invention the work of genius, and hence the proper subject matter of psychology rather than of logic? Can the process of arriving at an hypothesis (rather than of justifying it) be called inference or argument? Peirce does not lack historical precedent for such a view, of course, for induction has often been regarded, not as an evidencing process, but as a logic of discovery” (1946, p. 302). The self-conscious pursuit of the connection between the Case and the Rule of abduction — especially in the context of the connections between Rule and Result in deduction and Case and Rule in induction — calls for physicians (and others, such as detectives) to attend to phenomena more fully. Accomplished experienced physicians follow this “method” habitually — the “practical reason” of phronesis is, after all, “practiced” the way we practice a piano — but its self-conscious pursuit, we believe, makes it teachable as well.