Towards a genuinely medical model for psychiatric nosology This version revised March 21st, 2011 §* Nesse RM (1), *Stein DJ (2) (1) Department of Psychiatry, University of Michigan, East Hall Room 3018, Ann Arbor, MI 48109 USA (2) University of Cape Town *These authors contributed equally to this work § Corresponding author Email addresses: RMN: nesse@umich.edu DJS: Dan.Stein@uct.ac.za -1- Abstract Dissatisfaction with the current Diagnostic and Statistical Manual of Mental Disorders (DSMIV) is pervasive for a number of reasons.And although the pending DSM-5 will improve many details, it will not solve the overall problems. Here we discuss and suggest thatdissatisfaction with the DSM system results from unrealistic expectations arising from psychiatry’s commitment to an idealized version of the medical model in which the essence of each disorder is defined by its etiology. This is in contrast to the rest of medicine which recognizes many syndromes that arise from system failures that can have many causes. Further, we discuss how evolution of a more genuinely medical model for psychiatric nosology would offer a number of benefitsData demonstrating comorbidity, heterogeneity, and blurry boundaries for diagnostic categories that do not map neatly to specific pathophysiology do not demonstrate fatal flaws in the DSM, but instead challenge essentialist and reductionist assumptions, and encourage adoption of a model for understanding and classifying disorders more like that used in the rest of medicine -2- Background The current system of psychiatric diagnosis based on the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) [1] generates widespread criticism and intense controversy [2, 3]. Some clinicians, especially psychotherapists, ignore the DSM, saying that the detailed criteria do not help them to understand the origins of an individual’s symptoms. Others reify diagnostic categories as if they were natural kinds with “essences,” putting great effort into finding the best diagnostic code, at the expense of investigating the full range of a patient’s problems and possible causes [4]. Researchers protest requirements to use DSM categories that do not necessarily map well to neuroscience hypotheses [5, 6]. Members of the public are skeptical that disorders defined by committees are diseases, and that 50% of the population has a mental disorder. Physicians in other areas of medicine are understandably curious about why psychiatric diagnosis is so problematic. The current crisis in psychiatric nosology originated in the solution to a previous crisis. In the late 1970s, psychiatry awoke from a long dream to find itself floating on a couch in the backwaters of medicine. Emergency wake-up calls came from several directions. A 1973 article in Science reported that 12 “pseudopatients” who pretended to hear hallucinations were all admitted to hospitals and received a diagnosis of schizophrenia even though they acted normal after admission [7]. The episode was viewed as a humiliation for psychiatry, even though it is also easy to fake the symptoms of a heart attack or stroke [8]. Another wake-up call came from psychiatrists who were getting remarkable results from new medications, sometimes discharging patients who had been hospitalized for years [9]. -3- Psychopharmacologists soon demonstrated some specificity of drugs for different disorders, thus motivating searches for specific neurochemical abnormalities [10]. Drug companies began to investigate which drugs worked for which disorders, and regulatory agencies soon demanded objective diagnoses for efficacy studies. Insurance companies began reimbursing only for evidence-based treatment of bona fide medical disorders. Governments began funding neuroscience research on mental disorders. These dramatic changes brought psychiatry wide awake, and desperate to establish its medical status and scientific underpinnings. Studies in the 1970s soon documented pervasive problems with the DSM-II. For instance, UK and US approaches to schizophrenia diagnosis were found to differ dramatically [11]. The DSMII criteria for most diagnoses were markedly subjective. Consider the DSM-II definition for nonpsychotic depression: 300.4 Depressive neurosis: This disorder is manifested by an excessive reaction of depression due to an internal conflict or to an identifiable event such as the loss of a love object or cherished possession. It is to be distinguished from Involutional melancholia (q.v.) and Manic-depressive illness (q.v.). Reactive depressions or Depressive reactions are to be classified here.[12] Is depression after loss of a favorite cat “excessive?” One diagnostician would say, “Yes,” another, “Obviously not!” Such unreliable diagnoses made research impossible, and psychiatry's scientific pretensions laughable. Something had to be done. Objective diagnostic criteria were the solution. Like many advances, this one was proposed by apparent radicals who were at first ignored, then opposed, and finally lauded. In the early 1970s, psychiatrists in a weekly seminar at Washington University took up the challenge of creating -4- objective criteria. The resulting Research Diagnostic Criteria (RDC) defined the major recognized psychiatric disorders using checklists of observable characteristics that resembled a Chinese menu [13-15]. For instance, a diagnosis of depression required A, B, and C: A. Dysphoric mood characterized by symptoms such as the following: depressed, sad, blue, despondent, hopeless, "down in the dumps," irritable, fearful, worried, or discouraged. B. At least 5 of 8 symptoms/signs of depression, (e.g., weight loss , loss of energy, difficulty concentrating, suicidal thoughts or behavior) C. A duration of at least one month and no prior other condition that might cause such symptoms. This advance was not exactly greeted with hosannas. The first paper describing the RDC criteria [16], perhaps the most seminal in the recent history of psychiatry, was almost rejected for publication [13]. The editors accepted it on the condition that specific criteria be excluded from the text, but the authors slipped them back in as captions for the illustration. Very soon, the RDC were making new research possible. By the late 1970s, the inadequacy of the DSM-II was clear. The RDC criteria offered superior reliability, and they had support from researchers. Robert Spitzer, a major architect of the RDC, took charge of developing the DSM-III. Under his leadership, the psychiatric classification system eliminated theoretical constructs, such as “neurosis,” and replaced them with RDC-style checklists of observable signs and symptoms that defined specific entities. The goal was to make psychiatric diagnosis as scientific and objective as in the rest of medicine. -5- The DSM-III (Diagnostic and Statistical Manual, Third edition), published in 1980, listed criteria for 267 disorders [17]. It required diagnostic coding on five axes: I for the traditional main psychiatric disorders, II for personality disorders, III for medical conditions, IV for psychosocial factors, and V to reflect a Global Assessment of Functioning. DSM-III diagnoses were vastly more reliable. Researchers could finally compare treatments with confidence that patients in different studies actually had the same disorder. Neurobiological studies could proceed. Epidemiology became possible. The DSM-III made mental disorders seem much more like specific diseases. Psychiatrists could finally diagnose and treat specific conditions, just like other physicians. Subsequent revisions changed names and criteria, but all followed the same system of checklistsbased on observable indicators. The 1987 DSM-III-R eliminated some disorders, renamed others, and changed criteria. The 1994 DSM-IV used a more formal process and validation data to define 297 disorders in 886 pages [18]. It added requirements that most conditions be "clinically significant," to address concerns about excessive prevalence of many disorders [19]. The 2000 DSM-IV-TR kept almost all of the same entities, but revised the text descriptions and adjusted codes to match those of the International Classification of Diseases [1]. Progress Objective diagnostic criteria gave new life to psychiatric research. They made large-scale objective studies of psychiatric epidemiology possible for the first time. Standardized interviews with acronyms such as DIS, CIDI, SADS, and SCID, were developed to ensure that the same questions were asked in the same order, thus further increasing reliability, and making it possible for interviewers without clinical training to make psychiatric diagnoses. -6- Combined with modern survey sampling methods for ascertaining random population samples, these studies revealed a world of mental illness unimagined previously. We now can say with some confidence that about 1% of adults in the USA suffer from schizophrenia, and about 7% qualify for a diagnosis of depression in a given year. Despite the challenges of translation, we can also be somewhat confident that depression is about twice as prevalent in the USA as it is in most European countries [20], that anxiety and depressive disorders are about twice as common in women as men, and these differences are not due to differences in treatment seeking. Objective criteria also revealed remarkably high prevalence of alcohol related disorders; in the USA at the start of the 21st century the lifetime prevalence for alcohol abuse was 17.8%, for alcohol dependence, 12.5% [21]. Findings on specific disorders keep rolling in, but the most important discovery has been the gigantic health burden posed by mental disorders worldwide. Among all medical disorders for the world's population, mental disorders account for 10% of early death and disability, more than the total for cardiac disease [22]. In developed countries, the proportion of medical burden from mental disorders is over 15%, with about half caused by mood disorders. The higher proportions in developed societies are mainly because they have lower death rates from infectious disease [22]. Among other benefits, objective diagnostic criteria have made rigorous treatment studies possible. Double-blind placebo-controlled trials are now the norm, and randomized trials have been conducted for most disorders. Because participants are categorized using DSM criteria, researchers can have some confidence that results are generalizable to other patients who meet the same criteria. Since the publication of the DSM-III, single case studies of psychoanalysis for neurotic patients have been replaced by large studies of manualized psychotherapies for DSM -7- conditions. Clinical guidelines are now based on meta-analyses that integrate and summarize massive amounts of data [23]. In the clinic, DSM criteria have helped clinicians to identify subgroups of patients more accurately. For instance, obsessive-compulsive disorder (OCD) was hardly recognized 25 years ago. Now, thanks in part to specific DSM criteria, the diagnosis of OCD is better recognized, and more patients get appropriate specific treatment. Finally, the DSM system of diagnosis enhanced the stature of psychiatry in medicine. Specific categories with careful definitions made mental disorders seem much more like the diseases treated by other physicians. This decreased stigma, and this, in turn, facilitated patient consumer movements dedicated to improving recognition and treatment of mental disorders. The solution to the crisis of the 1970s has, in many respects succeeded beyond expectation. Problems Checklist systems for diagnoses have also revealed pervasive problems. We use the word "revealed" because we believe many of the problems associated with the DSM system were not caused by it, but were revealed by studies made possible by the objectivity of the DSM criteria. Several such problems deserve consideration. Political controversies make the newspapers. What kind of sexual behaviors are abnormal? Are emotional changes associated with menstruation a mental disorder? What about tobacco addiction? How severe do PTSD symptoms need to be to justify a clinical diagnosis and disability payments? Such controversies get disproportionate attention, and evidence rarely resolves them. They are nonetheless important, because diagnostic categories have real effects on -8- people’s lives and society; they influence whether particular behaviors are seen as harmless eccentricities, punishable crimes, or treatable disorders [24]. Huge prevalence rates for psychiatric disorders arouse skepticism. This has been a particular problem for depression. Lifetime rates based on the DSM-III-R were as high as 17% [25]. Tweaking the depression criteria for the DSM-IV brought rates down to 6.7% in the USA [26]. Where did all the people with major depression go? They are still there, but outside the official category, whose boundaries are obviously open to debate. Indeed, artificially sharp boundaries demarcate people with DSM-defined disorders from others. At the extremes, there is no difficulty. People with only a little social anxiety are normal. People so fearful they never leave their trailer at the end of a dead-end street have a serious problem. More than half of people experience distress when required to give a public presentation. Do they have social phobia? What about those of us who lose sleep the night before a big public presentation? The line separating normal from disorder is blurry and substantially arbitrary, very unlike the sharp division made by DSM criteria. Tweaking the criteria can adjust the percentage of people who have social phobia from 1% to 50%; the question is where to draw the line. This question is not simple to answer given that no “zone of rarity” separates those with mood or anxiety disorders from other people. A histogram of the severity of depression symptoms shows is no dip in the middle to suggest a natural breakpoint separating two groups [27-29]. Similarly, work on social phobia finds no particular number of symptoms that represents a discrete cutpoint for deciding whether or not a disorder is present [30]. -9- Heterogeneity is another problem. Individuals in the same diagnostic category may be very different. For instance, the diagnosis of major depression requires having five out of nine possible symptoms. One person may report depressed mood, weight loss, insomnia, fatigue and poor concentration, while another has decreased interest in activities, weight gain, sleeping too much, inappropriate guilt, and recurrent thoughts of death. Both individuals qualify for a diagnosis of major depression even though they have no symptoms in common! Some researchers hope to solve this problem by identifying diagnostic subcategories with more consistent clusters of symptoms and etiology. Others argue that excessive splitting causes high comorbidity. For instance, generalized anxiety disorder very often progresses to depression, and genetic vulnerability to these disorders is in part shared [31]. Massive comorbidity is perhaps the most troubling discovery from epidemiological studies, the best of which were called, appropriately enough, the "National Comorbidity Study (NCS) [32], " and the “NCS-Replication [26].” They found that most people who have one mental disorder have two or more additional diagnoses. For instance, more than half of individuals with OCD also qualify for a diagnosis of major depression. And, more than half of individuals with major depression qualify for diagnosis of anxiety disorder. The majority of individuals with alcohol dependence also qualify for a diagnosis of anxiety or mood disorder. These findings have frustrated the desire to identify discrete conditions with distinct etiologies [33-35] Reliability is vastly improved, but still relatively low for some conditions, especially those like schizophrenia and obsessive-compulsive disorder where superficially similar symptoms may occasionally be experienced by people who do not have a disorder. For all of the extraordinary efforts to operationalize criteria, and to embed them in standardized interviews, a number of DSM diagnoses have relatively poor reliability [36]. - 10 - Validity is the elephant in the closet. In the rest of medicine, disorders are defined by distinctive tissue pathology, but no biomarkers have been found that can validate DSM diagnoses for the major mental disorders. A pathologist can definitively confirm a diagnosis of Alzheimer's disease or multiple sclerosis, but bipolar disorder, schizophrenia, and autism have no known pathognomonic laboratory or autopsy findings. It is not for lack of looking! Exhaustive efforts have discovered some statistically significant average neurobiological differences in certain diagnostic groups (e.g. on brain imaging), but the findings are neither specific nor sensitive. Furthermore, brain-imaging abnormalities may reflect the activity of normal mechanisms that give rise to particular symptoms, rather than etiological factors. Neurogenetic research has emphasized that genes and environments interact to create vulnerability to disorder, but even the strongest specific gene-disorder correlations to date have been disappointingly weak [37, 38]. When the DSM-III was published, most expected that we would soon find distinctive pathophysiologies that would define particular diseases. Such pathognomonic changes have yet to be identified for any major psychiatric disorder. In conjunction with massive comorbidity, symptom heterogeneity and fuzzy boundaries, these problem have encouraged many to call for reexamining the conceptual foundations of psychiatric nosology. Inhibiting careful clinical assessment may be the most toxic clinical effect of checklist systems for diagnosing mental disorders [39, 40]. The DSM-III/IV criteria do not specifically argue that psychiatric disorders are categorical entities with clear boundaries and distinctive (or unitary underlying) causes. On the contrary, they emphasize that borders are fuzzy and causes multiple. To wit: “In DSM-IV, there is no assumption that each category of mental disorder is a - 11 - completely discrete entity with absolute boundaries dividing it from other mental disorders or from no mental disorder. There is also no assumption that all people described as having the same mental disorder are alike in all important ways [41] p xxii. Despite the avowedly atheoretical nature of the DSM, categories defined by checklists make diagnoses seem like discrete diseases or essential natural kinds [42]. This fosters a reductionist view that each has distinctive causes and the corresponding belief that it should be possible to find biomarkers for each disorder [43]. These tendencies foster neglect of rigorous historytaking and formulation of a comprehensive assessment of how multiple factors contribute to a particular patient’s symptoms [40, 44, 45] at the very moment when geneticists and neuroscientists are desperate for better phenomenology [39]. Conceptual challenges The above problems have led many leaders in psychiatry to conclude that the problems with psychiatric diagnosis are fundamental, and changes will require major reconceptualization. For instance, Allen Frances, chair of the DSM-IV Task Force, says, “We are at the epicycle stage of psychiatry where astronomy was before Copernicus and biology before Darwin. Our inelegant and complex current descriptive system will undoubtedly be replaced by explanatory knowledge that ties together the loose ends [46].” Steven Hyman, the former Director of NIMH says, “The problematic effects of diagnostic reification were revealed repeatedly in genetic studies, imaging studies, clinical trials, and types of studies where the rigid, operationalized criteria of the DSM-IV defined the goals of the investigation despite the fact that they appeared to be poor mirrors of nature” [47] p 731. - 12 - Thomas Insel, the current Director of NIMH calls for “reconceptualizing disorders of the mind as disorders of the brain” a process he says will transform psychiatry into “clinical neuroscience.” “With no validated biomarkers and too little in the way of novel medical treatments since 1980, … it is time to rethink mental disorders, recognizing that these are disorders of brain circuits...[48] p 1971.” “Our resources are more likely to be invested in a program to transform diagnosis by 2020, rather than modifying the current paradigm [49].” Frances is less optimistic about the short-term prospects for a successful revolution: “The DSMV goal to effect a “paradigm shift” in psychiatric diagnosis is absurdly premature… The incredible recent advances in neuroscience, molecular biology, and brain imaging that have taught us so much about normal brain functioning are still not relevant to the clinical practicalities of everyday psychiatric diagnosis. The clearest evidence supporting this disappointing fact is that not even one biological test is ready for inclusion in the criteria sets for DSM-V [50].” Concern about core conceptual issues is widely shared. The question is what to do about it. The criteria can be, and are being, modified. Are changes that are more substantial also possible? Proposed solutions DSM Revisions Many problems arose when the DSM-III system was implemented, so it is natural to hope that many can be solved by revising the criteria. This hope was confirmed to some extent by improvements in the DSM-IV. Revisions in DSM-5 will address many current problems. Experience with the DSM-IV,and new validation data give a solid foundation for these revisions. - 13 - Changes that are more radical are also being considered for DSM-5. For instance, the absence of a zone of rarity between normal problems and pathological conditions [27-29] has suggested that conditions such as anxiety, depression, and attention disorders would be better assessed as dimensions instead of as categories [51]. Dimensions more accurately represent clinical reality, but they pose practical and scientific obstacles. In practical terms, efficient communication demands words that refer to categories. In the rest of medicine, disorders such as hypertension are defined by cutoff values on dimensional scales. However, anxiety and depression are phenomenologically quite different from blood pressure. Blood pressure is an objectively measured number. In contrast, anxiety and depression scales sum up severity ratings for numerous symptoms. Thus, scores on such scales reflect the heterogeneity of symptoms as much as their severity. While offering advantages, dimensions are no more natural kinds for mental disorders than are diagnostic categories [52]. Diagnoses based on prototypes offer another possibility [53, 54]. Clinicians thinking about a disorder usually have in mind a prototype based on the findings in a cluster of typical cases. Patients similar to a prototypes are recognized as typical, those with different findings are viewed as atypical. Some have argued that concepts of "ideal types" will be helpful in reconceptualizing our nosology [42]. While operationalization poses challenges, a prototype approach to nosology matches human cognitive patterns, encourages attention to syndrome subtypes, discourages reification of categories, and encourages needed new attention to phenomenology. Whether such substantial changes should be implemented now has been the topic of many thoughtful commentaries about the trade-offs between the benefits of major changes for DSM-5 versus the costs of losing diagnostic consistency [3, 55]. For the most part, commentators - 14 - conclude that in the absence of definitive validation criteria, we are better off limiting changes to those that are especially well-justified [56]. Even those changes will require new clinical training, as well as extensive efforts to make prior and future research populations comparable. Replacing the DSM? Some thoughtful neuroscientists have encouraged different approaches, such as searching for “brain circuits” whose malfunctions could perhaps define specific disorders [57]. The notion of brain circuits reflects growing recognition that many functions are not localized, but distributed in "circuits" of neural pathways that connect diverse loci [2]. This calls attention to the need to attend to adaptive functions as well as structures. However, the analogy of brains with humandesigned circuits can mislead easily. Systems designed by engineers have discrete parts with specific functions and defined connections that make the system work. Evolved systems have components with indistinct boundaries, distributed functions, and innumerable connections very different from anything that an engineer could adequately describe [54, 58] . A related initiative encourages neuroscientists to look beyond DSM categories and classify patients using functional domains. Proposed Research Domain Criteria (RDoC) offer six domains: negative affect, positive affect, cognition, social processes, and arousal/regulatory processes, each of which has 2-6 subdomains [59]. These intersect with six units of analysis: genes, molecules, cells, circuits, behavior, and self-reports to create 36 (or more) variables. A study on anxiety disorders, for example, might examine people who show a heightened amygdala response to frightening pictures, regardless of whether their DSM diagnosis is panic disorder or social phobia. Another study might enroll subjects with a particular variation of the DISC1 gene, regardless of whether they have a diagnosis of bipolar disorder or schizophrenia - 15 - (both of which have been linked to DISC1). Such studies are argued to be essential for connecting the dots between biological abnormalities and the symptoms of mental illness [60]. This approach has the great benefit of trying to understand pathology in a framework based on normal functions, but it does not address directly the question of how to define specific disorders. It seems to be based on an idealized version of a medical model that assumes that each mental disorder corresponds to specific brain abnormalities. This will prove correct for some disorders. However, a model more like that used in the rest of medicine suggests the need to consider additional possibilities. Towards a genuinely medical model We are eager for the discovery of specific neurobiological abnormalities that cause major mental disorders. Such findings will provide a foundation for prevention and improved treatment for those disorders. However, the failure of such efforts to date encourages examination of the assumption that all mental disorders are discrete entities caused by specific identifiable neurobiological abnormalities. All behavior and emotion is mediated by brain mechanisms, of course. However, this does not imply that every disorder will have corresponding identifiable brain or circuit abnormalities [61, 62]. What are the alternatives? The medical model that guides thinking in other specialties calls attention to several. First, many medical conditions, such as fever, pain, and vomiting, are protective responses that are neither diseases nor specific symptoms of diseases. Second, many medical conditions are not diseases defined by a specific etiology, they are syndromes that arise - 16 - from failure of a particular functional system (or at a particular locus) that can be understood at the level of physiology without assuming that all cases arise from the same etiology. Third, many medical conditions, ranging from tinnitus to irritable bowel syndrome, are manifestations of functional failures that may or may not have specific etiologies. Finally, some mental disorders may differ in fundamental ways from other medical disorders, because they arise from abnormalities of information processing, and because they are manifested not in tissue pathology, but in behavior, emotion, and cognition. Emotions are defensive responses Physicians in other medical specialties routinely distinguish protective responses from direct manifestations of disease. Seizures, paralysis, and dyskinesias are direct manifestations of bodily defects. Fever, pain, and inflammation are protective responses shaped by natural selection, in conjunction with regulation systems that express them when the benefits are likely to exceed their costs [63]. Foreign material in the bronchi arouses cough, which expels it. Fever is useful during infections. Pain is useful when tissue is being damaged. Anxiety is useful in dangerous situations. Because they are painful and disabling, it is easy to assume that aversive emotions such as anxiety and low mood are maladaptive. However, like fever and pain, aversive emotions have offered selective advantages to our ancestors [64, 65]. This is confirmed by the existence of systems that regulate emotions; such systems could evolve only if emotions were useful in certain circumstances. Like pain and fever, emotions are pathological if they are expressed in inappropriate situations. The mechanisms that regulate expression seem especially vulnerable to - 17 - failure, based on high prevalence rates of chronic pain, chronic fatigue, anxiety disorders, and depression [63]. This has a profound implication for psychiatric diagnosis—whether an emotional response of a given intensity is normal or pathological can be determined only if you know what situation normally arouses the response, and whether or not that situation is present [24, 66, 67]. This conclusion informs one current debate about DSM-5. The diagnosis of major depression is based on the number, severity, and duration of symptoms, and the amount of associated disability. Information on life events is ignored, with one exception—in the two months after loss of a loved one, the diagnosis is excluded because symptoms of depression are thought to be normal. Wakefield and others have argued that other circumstances, such as being in intensive care, may also cause normal sadness [68]. They suggest modifying DSM criteria to avoid misdiagnosing such normal sadness as pathological depression. On the other side, Kendler and others have pointed out that the ICD has never had a grief exclusion, that depression arising from bereavement is clinically indistinguishable from other forms of depression, and that adding exclusions for other events would cause confusion and decrease reliability [69, 70]. They suggest omitting the grief criterion. The trade-offs are clear. Eliminating the grief exclusion would increase consistency and maintain reliability. Extending the exclusion to other situations that arouse extreme normal sadness would sacrifice reliability, but increase validity. It would also make DSM criteria more consistent with the rest of medicine, where all factors that could possibly arouse a defensive response are investigated before considering the possibility that the response itself is abnormal. The proposal to eliminate the grief exclusion illustrates the difference between psychiatry’s - 18 - idealized model, and a more pragmatic medical model based on understanding the adaptive functions of protective responses. Many conditions are syndromes, not diseases There is widespread agreement that psychiatric diagnoses should be based on etiology. However, many medical conditions are defined not by their etiology, but by syndromes that arise from failure of a functional system. For instance, congestive heart failure (CHF) results from inadequate cardiac output. It can have many causes, including fluid overload, valvular dysfunction, arrhythmia, and muscle abnormalities. Like many mental disorders, the manifestations of CHF are heterogeneous. For some patients edema is prominent, for others the main problem is fatigue, or dyspnea. This heterogeneity causes few concerns about the diagnosis of CHF, because the physiology that connects the functional failure to the findings is well understood. We suspect that many calls to define mental disorders based on etiology are not really demands to demonstrate necessary and sufficient distinctive causes. Finding a cause in the failure of some system would be most welcome. This has been proposed. For instance, deficient social capacities are thought by some to be the primary problem in autism, while problems in the development and selection of procedural strategies have been posited to underlie obsessivecompulsive disorder [71]. However, the challenges of understanding functional failures at higher levels of organization are substantial for behavioral systems, and somewhat different from those encountered in the rest of medicine. Physiology provides the rest of medicine with understanding at multiple levels for functions such as regulating blood pressure, acid-base balance, and glucose levels. For instance, - 19 - receptors in the Islets of Langerhans monitor glucose levels. If levels rise, they secrete insulin. If levels fall, alpha cells release glucagon. Scores of other mechanisms also come into play, but the system has discrete components with specific functions that collaborate to regulate a specific variable. The mechanisms that regulate behaviors related to social status are equally effective, but not susceptible to the same kind of analysis. Assessment of social status of self and others depends on processing innumerable cues in light of recalled prior information. Relative increases and decreases in status give rise to states that are difficult to characterize and measure. The resulting behaviors are even harder to quantify and study, because they depend on moment-to-moment variations in social interactions. A demeaning comment that begins to arouse resentment may be followed by a wink that changes the meaning from a threat into a shared joke. It would simplify the analysis if subsystems that regulate aspects of behavior could be mapped to specific brain loci or circuits. This is possible to some extent; the brain is not an amorphous organ whose components lack all functional specificity [72]. However, it is also nothing like a computer with anatomically discrete modules each with specific functions, and there is no reason to assume that mental syndromes are caused by abnormalities in components necessary to a particular function. For instance, even if the amygdala is required for experiencing anxiety (which may not be correct [73]), this does not imply that anxiety disorders arise from abnormalities in the amygdala. Panic attacks are associated with activation of the locus coeruleus, but this does not imply that they are caused by abnormalities in the locus coeruleus. As with many other syndromes in medicine, the causes can be many and distributed. Taking functional disorders seriously - 20 - The difficulty in finding specific pathophysiology associated with a disorder is by no means unique to psychiatry. Epilepsy and cardiac arrhythmias often have no identifiable etiologic lesions. However, both of these sets of disorders do have objective abnormal electrical patterns. Other disorders, such as irritable bowel syndrome and essential tremor, lack objective biomarkers, but are manifest in specific observable abnormalities. Other medical problems, such as tinnitus, dizziness, fatigue, headaches, and chronic pain, may have only subjective manifestations. Perhaps it is because they lack objective findings that these syndromes are often attributed to psychiatric causes. Disorders that disrupt function are hard to study if no specific pathophysiology can be identified. Furthermore, many have multiple possible causes. Tinnitus, for example, can result from causes as disparate as tumors, toxins, neuritis, and earwax. Many mental disorders are comparable, but with the functional failure arising in a behavioral or emotional system. Binge eating, temper explosions, and inability to stop repetitive behaviors may be examples. Like other disorders of function in medicine, individual instances may or may not have single specific causes. They are readily recognized and defined even without finding particular tissue abnormalities. Systems can fail because of regulation failures that cannot be fully understood from a reductionist perspective. Vicious cycles arising from positive feedback are central to an explanation of many conditions. Appendicitis, for instance, is initiated by inflammation that compromises circulation at the neck of the appendix. This decreases ability to control infection, resulting in more infection, causing more inflammation that further blocks circulation, until the appendix bursts. Pain caused by osteoporosis may limit exercise and result in additional bone loss. Obesity results in decreased exercise and increased obesity. Does dieting cause bingeing that arouses greater fear or obesity and more intense dieting? Does depressive withdrawal from - 21 - social life cause increased depression and further withdrawal? Does suspicion cause odd behavior which results in whispered gossip causing escalating suspicion, increasingly odd behavior, and further social isolation? This kind of cybernetic explanation is especially applicable to panic disorder. Patients with health concerns respond to slight changes in heart rate and breathing with fear that causes additional physiological arousal, which increases fear in a spiral that quickly escalates to a fullblown panic attack. Physiological variations explain the vulnerability of certain people to this spiral, and neural changes that mediate the attack, but a full explanation requires considering the positive feedback cycle at the level of cognition and emotion. Are mental disorders different? The term “mental disorders” is inherently misleading. Most people hear the phrase and contrast it with “physical disorders” that seem more real. Emphasizing that “mental disorders are brain disorders” helps to counter this misunderstanding, but at the cost of fostering the misconception that all mental disorders must be caused by brain abnormalities. As noted above, there are other possibilities. One rarely mentioned is the possibility that mental disorders could arise at the level of information processing, even if the brain is normal. Computer crashes are caused more often by software than hardware. If a program goes into an infinite loop or reaches a dead-end, it will crash, even if every chip and wire is intact. Do minds/brains also experience analogous failures? The analogy of minds with computers is imperfect. Software is designed by engineers who create modules with specific functions. This results in limited redundancy, so failure at any line of - 22 - code is likely to crash the program. Brains/minds are different. Because they were shaped by natural selection among miniscule changes over eons of time, they are not organized into such discrete modules, and they have myriad redundant interconnections, tested over time by natural selection, that result in remarkable resilience. That a child can grow up to function almost normally after early removal of an entire brain hemisphere [74] illustrates just how different brains are from computers. Nonetheless, some mental disorders may be best explained as failures of software, not hardware. A genuinely medical model for psychiatric nosology Specific neural pathophysiology will be discovered that that explains some major mental disorders, and these discoveries will lead to effective treatments for some devastating diseases. There should be no pause in pursing these discoveries. However, the assumption that all psychiatric disorders are discrete entities arising from specific brain abnormalities is not congruent with the disease models used in the rest of medicine. Other medical specialties recognize many problems that are not discrete entities defined by specific etiologies. Some, such as pain, fever, fatigue, and anxiety, are protective responses (that are susceptible to dysregulation). Other conditions are syndromes that can arise when a system is disrupted, from whatever cause. And some disorders do not have specific identifiable pathophysiology because they arise from dysfunctions in distributed systems, possibly because of information processing conflicts, or uncontrolled positive feedback. From this medical perspective, the DSM constellations recognized by experienced clinicians deserve respect. Their comorbidity, heterogeneity, unjustifiably sharp borders, and lack of specific identifiable pathophysiology are similar to what is observed for other medical conditions - 23 - such as CHF, epilepsy and tinnitus. The failure to define mutually exclusive categories for mental disorders each with a specific cause is not a flaw of the DSM, it is an indication that some conditions we are trying to understand are, like other medical disorders, syndromes with multiple possible etiologies. While the untidiness of DSM categories is unwelcome, it may accurately represent the structure of mental disorders. Nonetheless, revision is needed to correct some aspects of the DSM system that artificially separate psychiatry from the rest of medicine. In particular, psychiatric nosology could become much more like the rest of medicine if it sharply separated disease manifestations that are protective responses, such as fever, pain, anxiety and low mood, from direct manifestations of disease such as paralysis, cancer, hallucinations and dementia. Diagnosing an emotion as abnormal without carefully assessing the life situations that might arouse it is like diagnosing chronic pain without looking for possible causes of tissue damage. If psychiatric nosology were like that in the rest of medicine, decisions about the normality of an emotion would be based on knowledge of the situations that normally arouse the emotion, and the presence or absence of those situations. There are good reasons why this has not been done long ago. The first is that we are just now recognizing the adaptive significance of negative emotions and their comparability to other protective responses [75]. Another is that we still lack a full understanding of how emotions give selective advantages, and how the mind/brain processes information that indicates whether the current situation is one in which the response will likely be useful. Yet another is that our current environment is so different from that in which we evolved that many such responses are rarely useful now. Finally, there is the practical matter that reliability will be limited for any diagnostic criterion that requires assessing context. The same applies in the rest of medicine. A diagnosis of - 24 - "cough disorder" would be maximized by defining it as anything in excess of a specific number, severity, and duration of coughing episodes that occur in a specified timeframe. Instead of relying on such objective criteria, physicians assess whether a cough is a normal response to some other problem, or a result of an abnormality in the mechanism that regulates cough by investigating all possible causes of cough. Physicians do not demand that the cough regulation mechanism be abnormal to prescribe medications blocking the response. The “smoke detector principle” explains why it is usually safe to block aversive responses even when they are normal [63]. Another major change that would make psychiatric nosology more like that in the rest of medicine is to reconsider the nature of exclusion criteria in the DSM based on substance use and general medical conditions. In the DSM-IV, a diagnosis of major depression requires that: “The symptoms are not due to the direct physiological effects of a substance (e.g., a drug of abuse, a medication) or a general medical condition (e.g., hypothyroidism).” However, depression is still depression whether it arises mainly from genetic predisposition, major life failures, or interferon treatment. As with all medical syndromes, it is useful to distinguish different causes, and to recognize that they may well interact to cause an individual’s disorder. For instance, individuals with an s allele at the serotonin reuptake transporter promoter locus are more likely to develop depression in response to interferon treatment [76]. If the syndrome is the same as for depression induced by life situations and depression induced by interferon treatment, then they should be listed as subtypes. If the syndrome is different, as seems to be the case for schizophrenia vs. temporary psychotic symptoms induced by phencyclidine, then separate diagnoses may be justified. - 25 - A more genuinely medical model may offer additional suggestions for major changes in psychiatric nosology, perhaps including support for categories based on prototypes. Here, however, we do not attempt to lay out all the opportunities, but only to call attention to them. Conclusions Much of the dissatisfaction with current psychiatric nosology arises not from the inadequacies of the DSM system, but from unrealistic expectations aroused by an idealized model of diseases in which each is a natural kind defined by a specific etiology. This idealized model has practical consequences, including reification of psychiatric diagnoses in the clinic, and too optimistic a view of our ability to find neurobiological correlates of these conditions in the laboratory. Adoption of a more pragmatic model, of the kind that uses physiological/ecological understanding of adaptive systems to guide thinking in other medical specialties, would foster appreciation for the value of DSM categories despite their untidiness, and recognition that some changes are needed to make psychiatric diagnostic categories more like those in the rest of medicine. Competing interests None - 26 - Authors' contributions Both authors contributed equally to all aspects of preparing this manuscript. Acknowledgements We thank Brian Mickey for helpful comments on drafts of this manuscript. Bibliography 1. American Psychiatric Association: Diagnostic and statistical manual of mental disorders : DSM-IV-TR, 4th , text revision. edn. Washington, DC: American Psychiatric Association; 2000. 2. Akil H, Brenner S, Kandel E, Kendler KS, King MC, Scolnick E, Watson JD, Zoghbi HY: Medicine. The future of psychiatric research: genomes and neural circuits. Science 2010, 327(5973):1580-1581. 3. Hyman SE: The Diagnosis of Mental Disorders: The Problem of Reification. Ann Rev Clin Psych 2010, 6(1):155-179. 4. Luhrmann TM: Of Two Minds: An Anthropologist Looks at American Psychiatry. New York: Alfred A. Knopf; 2000. 5. van Praag HM: Nosologomania: a disorder of psychiatry. World Journal of Biological Psychiatry 2000, 1(3):151-158. 6. Hyman S: Can neuroscience be integrated into the DSM-V? Nature Reviews Neuroscience 2007, 8(9):725-732. 7. Rosenhan D: On being sane in insane places. Science 1973, 179(4070):250. - 27 - 8. Spitzer R: On pseudoscience in science, logic in remission, and psychiatric diagnosis: A critique of Rosenhan's" On being sane in insane places.". J Abnorm Psychol 1975, 84(5):442-452. 9. Delay J: Psychopharmacology and psychiatry. Presse Med 1967, 74(22):1151-1156. 10. Klein DF: Delineation of two drug-responsive anxiety syndromes. Psychopharmacology 1964, 5:397-408. 11. Wing JK: International comparisons in the study of the functional psychoses. Br Med Bull 1971, 27(1):77-81. 12. American Psychiatric Association. Committee on Nomenclature and Statistics.: Diagnostic and statistical manual of mental disorders, 2d edn. Washington DC: American Psychiatric Association; 1968. 13. Kendler KS, Munoz RA, Murphy G: The Development of the Feighner Criteria: A Historical Perspective. Am J Psychiatry 2010, 167(2):134-142. 14. Endicott J, Spitzer RL: A Diagnostic Interview: The Schedule for Affective Disorders and Schizophrenia. Arch Gen Psychiatry 1978, 35(7):837-844. 15. Feighner J, Robins E, Guze S, Woodruff Jr R, Winokur G, Munoz R: Diagnostic criteria for use in psychiatric research. Arch Gen Psych 1972, 26(1):57. 16. Spitzer R, Endicott J, Robins E: Research diagnostic criteria: rationale and reliability. Arch Gen Psych 1978, 35(6):773. 17. American Psychiatric Association: Diagnostic and statistical manual of mental disorders III, 3d edn. Washington, D.C.: American Psychiatric Association; 1980. 18. American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV). Washington D.C.: APA Press; 1994. - 28 - 19. Spitzer RL, Wakefield JC: DSM-IV diagnostic criterion for clinical significance: Does it help solve the false positive problem? Am J Psychiatry 1999, 156:1856-1864. 20. Demyttenaere K, Bruffaerts R, Posada-Villa J, et al.: Prevalence, severity, and unmet need for treatment of mental disorders in the World Health Organization World Mental Health Surveys. JAMA 2004, 291:2581-2590. 21. Hasin DS, Stinson FS, Ogburn E, Grant BF: Prevalence, correlates, disability, and comorbidity of DSM-IV alcohol abuse and dependence in the United States: results from the National Epidemiologic Survey on Alcohol and Related Conditions. Arch Gen Psychiatry 2007, 64(7):830-842. 22. Lopez AD: Global burden of disease and risk factors: Oxford University Press, USA; 2006. 23. Rush AJ: Clinical practice guidelines: Good news, bad news, or no news? Arch Gen Psych 1993:483-490. 24. Stein D, Phillips K, Bolton D, Fulford K, Sadler J, Kendler K: What is a mental/psychiatric disorder? From DSM-IV to DSM-V. Psychological Medicine 2010, 40(11):1759-1765. 25. Kessler RC, McGonagle KC, Zhao S, Nelson C, Hughes M, Eshleman S, Wittchen HU, Kendler KS: Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity Survey. Arch Gen Psych 1994, 51:8-19. 26. Kessler RC, Chiu WT, Demler O, Walters EE: Prevalence, severity, and comorbidity of twelve-month DSM-IV disorders in the National Comorbidity Survey Replication (NCS-R). Arch Gen Psych 2005, 62(6):617. - 29 - 27. Ruscio J, Zimmerman M, McGlinchey JB, Chelminski I, Young D: Diagnosing major depressive disorder XI: a taxometric investigation of the structure underlying DSMIV symptoms. J Nerv Ment Dis 2007, 195(1):10-19. 28. Slade T, Andrews G: Latent structure of depression in a community sample: a taxometric analysis. Psychol Med 2005, 35(4):489-497. 29. Kendell R, Jablensky A: Distinguishing Between the Validity and Utility of Psychiatric Diagnoses. Am J Psychiatry 2003, 160(1):4-12. 30. DJ S, AM R, S L, M P, J A, LH A, C B, E B, K D, S F et al: Subtyping social anxiety disorder in developed and developing countries. Depression and Anxiety 2010, 27(4):390-403. 31. Kendler KS, Neale MC, Kessler RC, Heath AC, Eaves LJ: Major depression and generalized anxiety disorder. Same genes, (partly) different environments? Arch Gen Psych 1992, 49(9):716-722. 32. Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshelman S, Willchen HU, Kendler KS: Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the national comorbidity survey. Arch Gen Psych 1994, 51:8-19. 33. Wittchen HU, Zhao S, Kessler RC, et al.: DSM-III-R generalized anxiety disorder in the National Comorbidity Survey. Arch Gen Psych 1994, 51:355-364. 34. Kessler RC, Sonnega A, Bromet E, Hughes ML, CB N: Posttraumatic stress disorder in the National Comorbidity Survey. Arch Gen Psych 1995, 52:1048-1060. - 30 - 35. Kessler RC, Stang P, Wittchen HU, Stein MR, Walters EE: Lifetime co-morbidities between social phobia and mood disorders in the US National Comorbidity Survey. Psychol Med 1999, 29:555-567. 36. Williams JB, Gibbon M, First MB, Spitzer RL, Davies M, Borus J, Howes MJ, Kane J, Pope HG, Jr., Rounsaville B et al: The Structured Clinical Interview for DSM-III-R (SCID). II. Multisite test-retest reliability. Arch Gen Psychiatry 1992, 49(8):630-636. 37. Kendler KS: Reflections on the Relationship Between Psychiatric Genetics and Psychiatric Nosology. Am J Psychiatry 2006, 163(7):1138-1146. 38. Karg K, Burmeister M, Shedden K, Sen S: The Serotonin Transporter Promoter Variant (5-HTTLPR), Stress, and Depression Meta-analysis Revisited: Evidence of Genetic Moderation. Arch Gen Psych 2011:archgenpsychiatry. 2010.2189 v2011. 39. Kendler KS, Zachar P: The incredible insecurity of psychiatric nosology. Philosophical issues in psychiatry: Explanation, phenomenology and nosology 2008:370–385. 40. Andreasen NC: DSM and the death of phenomenology in America: an example of unintended consequences. Schizophrenia Bulletin 2007, 33(1):108. 41. American Psychiatric Association: Diagnostic and statistical manual of mental disorders : DSM-IV, 4th edn. Washington, DC: American Psychiatric Association; 1994. 42. Ghaemi SN: Nosologomania: DSM & Karl Jaspers' Critique of Kraepelin. Philosophy, Ethics, and Humanities in Medicine 2009, 4(1):10. 43. Goldberg D: Plato versus Aristotle: categorical and dimensional models for common mental disorders. Comprehensive Psychiatry 2000, 41(2 Suppl 1):8-13. - 31 - 44. Schwartz MA, Wiggins OP: Diagnosis and ideal types: A contribution to psychiatric classification* 1. Comprehensive Psychiatry 1987, 28(4):277-291. 45. Schwartz MA, Wiggins OP: Systems and the structuring of meaning: contributions to a biopsychosocial medicine. Am J Psychiatry 1986, 143(10):1213. 46. Frances AJF, Egger HL: Whither psychiatric diagnosis. Aust N Z J Psychiatry 2003, 33:161-165. 47. Miller G: Beyond DSM: Seeking a Brain-Based Classification of Mental Illness. Science 2010, 327(5972):1437. 48. Insel TR, Wang PS: Rethinking Mental Illness. JAMA: The Journal of the American Medical Association 2010, 303(19):1970-1971. 49. Greenberg G: Inside the battle to define mental illness. Wired 2010(Jan 2010). 50. Frances A: A Warning Sign on the Road to DSM-V: Beware of Its Unintended Consequences Psychiatric Times 2010, 26(8). 51. Helzer JE: A Proposal for Incorporating Clinically Relevant Dimensions Into DSM5. In: The Conceptual Evolution of DSM-5. edn. Edited by Regier DA, Narrow WE, Kuhl EA. Washington DC: American Psychiatric Publisher; 2010: 81-96. 52. Kessler RC: The categorical versus dimensional assessment controversy in the sociology of mental illness. Journal of Health and Social Behavior 2002, 43(2):171-188. 53. Westen D, Heim A, Morrison K, Patterson M, Campbell L: Simplifying diagnosis using a prototype-matching approach: Implications for the next edition of the DSM. Rethinking the DSM: A psychological perspective 2002:221–250. 54. Stein DJ: The Philosophy of Psychopharmacology: Smart Pills, Happy Pills, Pep Pills. Cambridge: Cambridge University Press; 2008. - 32 - 55. First MB, Pincus HA, Levine JB, Williams JBW, Ustun B, Peele R: Clinical Utility as a Criterion for Revising Psychiatric Diagnoses. Am J Psychiatry 2004, 161(6):946-954. 56. Kendler K: Alternative futures for the DSM revision process: iteration v. paradigm shift. The British Journal of Psychiatry 2010, 197(4):263. 57. Insel TR, Quirion R: Psychiatry as a clinical neuroscience discipline. JAMA 2005, 294(17):2221. 58. Nesse RM: Maladaptation and natural selection. Q Rev Biol 2005, 80(1):62-70. 59. Insel T, Cuthbert B, Garvey M, Heinssen R, Pine D, Quinn K, Sanislow C, Wang P: Research Domain Criteria (RDoC): Toward a new classification framework for research on mental disorders. Am J Psychiatry 2010, 167(7):748. 60. Cuthbert B, Insel T: The data of diagnosis: new approaches to psychiatric classification. Psychiatry 2010, 73(4):311-314. 61. Zachar P, Kendler KS: Psychiatric disorders: A conceptual taxonomy. Am J Psychiatry 2007, 164:557-565. 62. Kendler KS: Explanatory Models for Psychiatric Illness. Am J Psychiatry 2008, 165(6):695-702. 63. Nesse RM: Natural Selection and the Regulation of Defenses: A Signal Detection Analysis of the Smoke Detector Principle. Evolution and Human Behavior 2005, 26:88-105. 64. Nesse RM: Emotional Disorders in Evolutionary Perspective. Br J Med Psychol 1998, 71(4):397-416. - 33 - 65. Nesse RM: Evolutionary origins and functions of emotions. In: The Oxford Companion to Emotion and the Affective Sciences. edn. Edited by Scherer K, Sander D. Oxford: Oxford University Press; 2009: 159-164. 66. Nesse RM: Explaining depression: Neuroscience is not enough, evolution is essential. In: Understanding depression: A translational approach. edn. Edited by Pariente CM, Nesse RM, Nutt DJ, Wolpert L. Oxford, UK: Oxford University Press; 2009: 17-36. 67. Nesse RM: On the difficulty of defining disease: a Darwinian perspective. Medical Health Care and Philosophy 2001, 4(1):37-46. 68. Wakefield J, Schmitz M, Baer J: Does the DSM-IV clinical significance criterion for major depression reduce false positives? Evidence from the National Comorbidity Survey Replication. Am J Psychiatry 2010, 167(3):298. 69. Zisook S, Kendler KS: Is bereavement-related depression different than nonbereavement-related depression? Psychol Med 2007, 37(06):779-794. 70. Kendler KS: DSM-5 Workgroup manuscript. 2010. 71. Stein DJ: Cognitive-Affective Neuroscience of Mood and Anxiety Disorders. In. London: Martin Dunitz; 2003. 72. Phan KL, Wager TD, Taylor SF, Liberzon I: Functional neuroimaging studies of human emotions. CNS spectrums 2004, 9:258-266. 73. Bach DR, Talmi D, Hurlemann R, Patin A, Dolan RJ: Automatic relevance detection in the absence of a functional amygdala. Neuropsychologia, In Press, Corrected Proof. 74. Vining EP, Freeman JM, Pillas DJ, Uematsu S, Carson BS, Brandt J, Boatman D, Pulsifer MB, Zuckerberg A: Why would you remove half a brain? The outcome of 58 - 34 - children after hemispherectomy-the Johns Hopkins experience: 1968 to 1996. Pediatrics 1997, 100(2 Pt 1):163-171. 75. Nesse RM, Ellsworth PC: Evolution, emotions, and emotional disorders. Am Psychol 2009, 64(2):129-139. 76. Lotrich FE, Ferrell RE, Rabinovitz M, Pollock BG: Risk for Depression During Interferon-Alpha Treatment Is Affected by the Serotonin Transporter Polymorphism. Biological Psychiatry 2009, 65(4):344-348. - 35 -