Article Text

Download PDFPDF

The definition of mental disorder: evolving but dysfunctional?
  1. Rachel Bingham1,
  2. Natalie Banner2
  1. 1General Practice Speciality Training Scheme, University College Hospital London, London, UK
  2. 2Wellcome Trust, London, UK
  1. Correspondence to Dr Rachel Bingham, General Practice Speciality Training Scheme, University College Hospital London, 250 Euston Road, London NW1 2PG, UK; rachelbingham{at}


Extensive and diverse conceptual work towards developing a definition of ‘mental disorder’ was motivated by the declassification of homosexuality from the Diagnostic and Statistical Manual in 1973. This highly politicised event was understood as a call for psychiatry to provide assurances against further misclassification on the basis of discrimination or socio-political deviance. Today, if a definition of mental disorder fails to exclude homosexuality, then it fails to provide this safeguard against potential abuses and therefore fails to do an important part of the work it was intended to do. We argue that fact-based definitions of mental disorder, relying on scientific theory, fail to offer a robust definition of mental disorder that excludes homosexuality. Definitions of mental disorder based on values do not fare better: these definitions are silent on questions about the diagnostic status of individuals in oppressive societies and over-inclusive of mental or behavioural states that happen to be negatively valued in the individual's social context. We consider the latest definition proposed for the Diagnostic and Statistical Manual-5 (DSM-5) in light of these observations. We argue that definition fails to improve on these earlier deficiencies. Its inclusion in the manual may offer false reassurance against repetition of past misclassifications. We conclude with a provocation that if candidate definitions of mental disorder are unable to exclude homosexuality, it might perhaps be preferable not to attempt a definition at all.

  • Psychiatry
  • Concept of Mental Health
View Full Text

Statistics from


Successive editions of the Diagnostic and Statistical Manual (DSM) triggered debate not only on what the manual says about the various conditions included, but also about what the manual says about mental disorder itself. At the dawn of publication of the latest edition of the DSM (DSM-5), the question of whether and how the DSM should define mental disorder remains as controversial as ever. This paper revisits the origins of this debate: the need, in the first place, to employ a definition of disorder and the work that it is hoped a definition may do.

A pivotal moment in the history of psychiatric nosology was arguably the declassification of homosexuality in the DSM. This raised allegations of political biases in diagnosis and shifted onto psychiatry the burden of accounting for the kind of conditions that psychiatrists treat. This motivated a great deal of conceptual work. Taking this historic moment as our reference point, we argue that existing definitions of disorder have failed to provide a robust conceptual account of the exclusion of homosexuality as a disorder. Rather, it is widely acknowledged that in the case of declassification from the DSM, socio-political elements took the day. Yet, if available definitions do not provide a satisfactory account in this pivotal case, are they doing the work they are designed to do? If not, might the false reassurance of a ‘definition of disorder’ leave us more vulnerable to repetition of past misclassifications—perhaps of other non-pathological states—than if the DSM was to provide no definition at all?

Homosexuality and the diagnostic manuals

Reflecting the diagnostic manual's historical context, homosexuality appeared in the first DSM in 1952, among ‘sociopathic personality disorders’. It was re-categorised with ‘sexual deviations’ in DSM-II in 1968. In 1973, following years of debate, protests and ultimately a vote by the Board of Trustees, homosexuality was removed from the classification. Related scientific theory was of course much discussed. However, it is widely accepted that ultimately the removal of homosexuality was not so much an outcome of new scientific knowledge, as, as Bayer put it, ‘an action demanded by the ideological temper of the times’.1

Defenders and critics of psychiatric nosology alike understood declassification of homosexuality to be a critical moment in the history of psychiatric nosology and of psychiatry itself. For critics, as Kutchins and Kirck put it, The best way to appreciate how mental disorders are invented is to understand how one diagnosis, homosexuality, was expunged from DSM.2

Those interested in explaining and refining psychiatry's conceptual apparatus were also motivated by the need to respond to what Wakefield describes as: Public concerns about misapplication of the term disorder [which] underlie accusations of sexual, racial, and sexual orientational biases in diagnosis.3

In other words, expert consensus over the legitimate conceptual domain of psychiatry was required to account for the exclusion of high-profile, apparently politically driven, former inclusions. To do so, the definition would need to successfully distinguish conditions that are disorders from those that are not.

Spitzer, likewise, expected that a definition would do a significant amount of work: When I defended the decision to include in DSM-III and DSMIII-R a definition of mental disorder, I argued that the definition was often helpful in resolving diagnostic controversies (e.g., Is homosexuality itself a disorder? Is premenstrual dysphoric disorder a disorder?).4

Indeed, the foreword to DSM-IV stated that a definition of mental disorder is included “to guide decisions regarding which conditions on the boundary between normality and pathology should be included.”5

Accordingly, following the declassification of homosexuality, a robust definition of disorder was considered necessary to re-establish the legitimacy of psychiatry as a scientific clinical discipline and to address (and safeguard against repetition of) past abuses. To do so, the definition should distinguish conditions that are not illnesses, from those that are, by implication, the legitimate target of psychiatric intervention.

The litmus test

We propose that if it is correct that homosexuality is among the most historically significant exclusions from the DSM, and that this event motivated those concerned to provide a definition of mental disorder for the DSM, it is reasonable to expect that if a candidate definition of mental disorder is put forward, it should be able to exclude homosexuality. It may be objected that homosexuality is not analogous to genuine mental disorders because of one of many plausible differences. For example, perhaps homosexuality is not like schizophrenia because there is nothing intrinsic to homosexuality that causes distress or impairment. These are precisely the sorts of distinctions that a definition of disorder hopes to capture, and that are discussed below.

In the following, we test various influential definitions of mental disorder by their ability to exclude homosexuality. This is by no means an exhaustive list, but is rather illustrative of the kinds of approaches that have been taken to defining mental disorder over the past 40 years. We suggest that this analysis reveals weaknesses across the range of putative definitions generated by diverse efforts towards providing a definition of disorder. This challenge suggests that these efforts are themselves undermined by the inability to exclude this important area from the domain of psychiatry. If correct, we argue that the usefulness of definition itself is called into question.

We take as our first premise that homosexuality should be excluded by any useful definition of disorder. We assume without discussion that homosexuality is not a disorderi and perceive this to be the consensus among those concerned to delineate the legitimate domain of psychiatry. For the current purpose, as argued above, because a great deal of work undertaken towards providing a definition of mental disorder was partly motivated by the need to account for the reclassification of homosexuality, success in this area appears a reasonable standard. Our concern, however, is not only whether homosexuality is excluded by the definition now in a society that broadly recognises that it is not pathological, but whether it would be excluded in a society that does not recognise this. This is important not solely to exclude homosexuality, but to ensure that we have the sort of definition that is able to exclude—in general—states that are wrongly pathologised due to prejudice and social exclusion, including those we might not have identified as such yet. Accordingly, while an important reminder of the dangers of over-diagnosis, we do not consider sufficient, in itself, the DSM's stipulation that, “Socially deviant behavior (e.g., political, religious, or sexual) and conflicts that are primarily between the individual and society are not mental disorders unless the deviance or conflict results from a dysfunction in the individual”.

Now, the above exclusionary statement does exclude homosexuality in a society that considers problems associated with homosexuality to arise primarily from conflict between the individual and the society, rather than from an underlying dysfunction. However, this understanding resulted from social change (including de-pathologisation of homosexuality) rather than from some finding about the underlying cause of homosexuality, as we further discuss below. Thus, ultimately, the exclusion of social deviance is only useful insofar as we are able to recognise it as such and does not help in cases where a dysfunction is thought to be present. In other words, even where the exclusion criterion applies, we rely on the absence of a dysfunction. This is explored further in the following section.

Definitions of disorder

Across the board, definitions of disorder comprise one or both of a set of ‘fact-based’ criteria and/or a set of ‘value-based’ criteria. Fact-based criteria describe scientific facts, including biomedical and statistical information, while value-based criteria specify that to qualify as a disorder a state must be, for example, considered harmful or distressing or otherwise negatively valued. The validity of the fact-value distinction is of course widely disputed as it is a matter of significant debate as to whether facts can be value-free. This will not be discussed further as our concern is not so much with the validity of the distinction as it is with the work done by the criteria generated on both sides. We consider first the role of fact-based criteria in accounts of disorder, and second the value-based criteria: while we recognise that these cannot always be readily divided, conceptually separating these two kinds of criteria is instructive for an analysis of the grounds on which homosexuality may or may not be excluded from definitions of disorder.

‘Fact-based’ criteria

Scientific and technological advancements have led to immense advances in knowledge of normal physiological functioning and disease aetiology. In somatic medicine, some disorders are diagnosed on the basis of discovering an underlying somatic lesion and it appears reasonable to say that in these cases a lesion model successfully delineates disorder from non-disorder. Branches of biological psychiatry remain optimistic about the delivery of comparable insights through technologies including functional MRI and genetic analysis (see, eg, Insel and Wang6).

However, even perfectly described neuropsychological states may prove a disappointing means of defining disorder: uncovering the neurobiological correlates of a given psychiatric diagnosis may assist understanding of aetiology and potential interventions but will not confirm that the diagnosis reflects a true mental disorder; this is already decided on the basis of the symptoms and behaviours presented. For example, a scientifically robust description of the area in the brain governing sexuality would not permit the inference that homosexuality is a disorder any more than description of the neural correlates of falling in love or criminality would make these mental disorders. Identifying the neural or genetic basis for a set of behaviours or symptoms does not in itself pick out what is pathological or disordered. Likewise a ‘gay gene’ would not justify restoring homosexuality to the DSM because this alone would not identify anything pathological in the genetic makeup of people identifying as homosexual. Indeed, some gay rights activists have advocated for research into biological and genetic determinants of homosexuality because of perceived advantages in reduction of blame and stigma associated with the claim that homosexuality is genetically influenced and is not a lifestyle choice.7

Recognition of the limitations of ‘lesion’ concepts of disease has meant that the major contenders among ‘fact-based’ criteria attempt to define disorder as deviation from a defined norm. This may be defined statistically, or by a normative theory of function and dysfunction, usually couched in evolutionary terms.

The most prominent proponent of a biostatistical theory, Boorse,8 defines disease (which he considers to be a solely factual, non-evaluative matter) in terms of statistical deviation in functioning from a defined population norm. It is imperative to the theory that an accurate account of what constitutes normal functioning can be given, otherwise it will fail to distinguish disease from rare, negatively valued states that are not disease (such as homosexuality; UK estimates suggest approximately 1% of the UK population identify as gay or lesbian) and at the other extreme for failing to recognise common diseases. On Boorse's view, a deviation from normal functioning is one that confers a reproductive disadvantage: taken narrowly (and we highlight alternative views in the following section), the inability to procreate through homosexual relationships does confer a reproductive disadvantage and is therefore construed as a statistical deviation from normal functioning. Thus homosexuality may be included as a disorder in a biostatistical theory, as Boorse himself accepts.9

Theories of statistical deviation rely on a definition of ‘function’ in order to track the right conditions, and indeed the DSM-IV relied on the notion of ‘dysfunction’ in its definition of disorder: Whatever [the condition's] original cause, it must currently be considered a manifestation of a behavioural, psychological or biological dysfunction in the individual. Neither deviant behaviour (e.g., political, religious or sexual) nor conflicts that are primarily between the individual and society are mental disorders unless the deviance or conflict is a symptom of a dysfunction in the individual as described above.5

A scientific theory of dysfunction therefore has a great deal of work to do. Arguably the most influential attempt to develop a credible theory of dysfunction is Wakefield's Harmful Dysfunction Analysis. This was described by Spitzer as “as a major conceptual advance over previous attempts” to define disorder, including Spitzer's own.4

Wakefield's analysis of ‘dysfunction’ begins with ‘function’. This refers to “why the mechanism exists or why it has the form that it does” and is filled out by evolutionary theory.3 For example, “it is the pumping [of the heart], and not the sound, that explains why we have hearts and why hearts are structured as they are”.3 Wakefield argues that evolutionary theory also explains mental functions: Perception, language, learning... and all the other furniture of the mind have their distinctive functions that explain why they exist in the first place and why they have the structures that they do.3

This is proposed as a ‘purely scientific concept’.3 Yet, there is a profound equivocation on which aspect of the definition (the science, or the harm) is supposed to provide the exclusion for homosexuality. It is important to note that on Wakefield's definition a disorder must be harmful and dysfunctional; for a condition not to be a disorder, therefore, it must either be not harmful or not dysfunctional, at least one of these negative criteria must be fulfilled to exclude the condition from being considered a disorder. On the one hand, for Wakefield, The requirement that a disorder must involve a dysfunction places severe constraints on which negative conditions can be considered disorders and thus protects against arbitrary labelling of socially disvalued conditions as disorders.3

But, on the other hand, in the same paper: [V]alue differences, rather than any dispute over facts, may be what makes some diagnostic controversies, such as that over the pathological status of homosexuality, so intractable.3

Spitzer argued that homosexuality should be removed from the DSM following his observations that neither distress nor impairment of function was entailed. This followed his contact with a prominent and outspoken group of happy, well-adjusted homosexuals who argued that their sexuality had little bearing on with their ability to function at a high level.1 He therefore attributes the exclusion of homosexuality to values: [E]ven if there is a dysfunction in some forms of homosexuality, the controversy about homosexuality may ultimately be about values—in Wakefield’s terms, a question of whether the dysfunction (assuming there is one) is a harmful or negative condition.4

The equivocation over whether homosexuality is excluded by facts or values occurs with good reason as it is far from obvious that ‘dysfunction’—defined as above—excludes homosexuality. The question of the evolutionary adaptiveness of homosexuality is comparable to theories of ‘biological advantage’, in which disorder is defined as a condition that confers a reproductive disadvantage or early mortality.10 This appears not protective against past misdiagnosis, but rather reminiscent of it: for example, the original inclusion of homosexuality under psychiatric nosology, as opposed to a ‘moral’ or legal category, arguably stemmed from eugenic assumptions of early evolutionary theory; ‘degeneration theory’ in particular.11

It may be objected that only early, discredited, understandings of evolutionary theory see homosexuality as a dysfunction in evolutionary terms. Recent theories argue for homosexuality as an adaptive trait. Kirkpatrick gives the following examples12: the ‘kin-selection hypothesis’ explains homosexuality in evolutionary terms as an altruistic foregoing of reproduction conducive to raising relatives’ offspring. The ‘parental-manipulation hypothesis’ explains homosexuality as selection by the parents of only some offspring to reproduce in order to confer greater advantage upon fewer grandchildren. Alternatively, ‘balanced polymorphism’ links inheritance of homosexuality to a second, positive trait. Accepting one of these theories would exclude homosexuality from diagnosis using the harmful dysfunction analysis.

However, the diversity of explanations emphasises that once we reject straightforward reproductive advantage, evolutionary psychology deals with relatively weak hypotheses. Within discussion of the Harmful Dysfunction Analysis alone, a mind-boggling variety of hypotheses are discussed: depressive symptoms are adaptive in promoting withdrawal from adverse environments13; manipulativeness, aggressiveness and deceptiveness are predatory strategies13; personality disorder arises when adaptive mechanisms allow early adverse environments to shape personality14; PTSD is an ‘adaptive’ response to trauma14; ‘paraphilic rapism’ confers ‘reproductive advantage in the extreme’.15 At least some evolutionary theoretic hypotheses border on the wildly speculative. This may not undermine evolutionary psychology as a valid and fascinating project, but it does not provide a robust account of the definition of mental disorder, nor qualify the status of homosexuality as either within or beyond the domain of psychiatry.


If we are correct that fact-based criteria for mental disorder do not exclude homosexuality, then we depend on value-based criteria to do the work of exclusion. We have argued that the major definitions of disorder may allow homosexuality to be characterised, on their own terms, as factually different to heterosexuality (eg, as a difference in the brain, or a reproductive disadvantage, or as statistically relatively infrequent). If so, then the only barrier these definitions provide against classification of homosexuality as a ‘disorder’ is the criterion that to be a disorder the state in question must cause harm or be negatively valued.

The problem is that many societies have conceived of homosexuality as a negatively valued state, so a robust account of what kind of negatively valued state counts in the definition of disorder is clearly required. On Wakefield's account, homosexuality would only be explicitly excluded from the classification system because it is not perceived as harmful. On the one hand, it seems clear that the evaluation of harm cannot simply be based on the standards of the individual's society, as this suggests that a gay individual in an oppressive society might be legitimately classified as having a disorder. Wakefield's position, however, appears to commit him to just such a claim. ‘Harm’ is defined, ‘as judged by the standards of the person’s culture’.3 “[H]armful” reflects social, not individual, values. For example, in a literate society, a person who does not value reading still has a dyslexic disorder if incapable of learning to read due to a brain dysfunction; and, in a society valuing reproductive capacity, a sterile individual has a disorder even if he or she does not want children.16

Spitzer appears to agree: Because pedophilic behavior results in the victimization of children, the dysfunction also represents a harmful condition by social standards, Thus pedophilia (at least when severe), is correctly classified as a disorder, not a normal variant.4

Homosexuality arguably only becomes a problem for individuals because of societal prejudice and discrimination; there is nothing intrinsically harmful about homosexuality (and it does not cause harm to other people). Could proponents of value-based criteria exclude homosexuality (and all diagnoses that represent only prejudice or social/political deviance) by saying more about what constitutes the relevant type of ‘harm’? The problem here is that because harm is conceived for Wakefield and others as a purely evaluative term, by definition there is no more to say than what is valued. This is not to say ‘anything goes’, but rather, there is no exclusion of conditions that are only harmful because of social, political or cultural values. Moreover, to try to exclude conditions that are only harmful in this respect would also exclude conditions that are considered to be disorders, as in Wakefield's examples of dyslexia and infertility.

In any case, it would be an artificial polarisation to attempt to understand distress or harm as separable from cultural values, and it is not obviously desirable for healthcare to be informed by a narrow view of harm. This would run counter to ideals of holistic healthcare and attempts understand people in context. If correct, Wakefield rightly extends ‘harm’ to cultural standards. But if so, then socially undesirable conditions are not excluded from the definition of disorder by distress or harm criteria.

Elsewhere, work on evaluative aspects of mental disorder extending beyond conceptual analysis has offered greater insight into the role of values. For example, work emphasising practitioners’ need to be aware of their own values offers a potential safeguard against future abuses of diagnosis (eg, Fulford's17). Bolton's study emphasising distress and disability prioritises the experiences and needs of the individual over apparently scientific criteria for diagnosis.18 Of note, these projects do not attempt to generate concrete definitions of ‘disorder’, but rather seek to understand the processes involved in diagnosis including the influences of various stakeholders. Thus, while offering a more nuanced understanding of the multiple influences on diagnosis, these projects do not provide a ‘definition’ of disorder or conceptual means of excluding homosexuality. They do, however, offer insights into why homosexuality may have previously been included and hope of greater practitioners’ awareness of their own impact. Ultimately it is here that potential to safeguard against future misclassifications and abuses may lie: without the expectation that conceptual analysis would do this work.

The foregoing suggests that efforts to define mental disorder have so far failed to satisfy their own central objectives: the need to account for the wrongful inclusion of homosexuality in earlier manuals and the need to provide safeguards against future similar abuses. Where does this leave us today, particularly with respect to the possibilities of generating a definition for the DSM? In the following, we bring these considerations into the present with a brief look at the definition of disorder now employed by DSM-5.

What next for the definition of disorder?

The definition in DSM-5 differs in subtle but significant ways from its predecessor: A mental disorder is a syndrome characterized by clinically significant disturbance in an individual's cognition, emotion regulation, or behavior that reflects a dysfunction in the psychological, biological, or developmental processes underlying mental functioning. Mental disorders are usually associated with significant distress or disability in social, occupational, or other important activities. An expectable or culturally approved response to a common stressor or loss, such as the death of a loved one, is not a mental disorder. Socially deviant behavior (e.g., political, religious, or sexual) and conflicts that are primarily between the individual and society are not mental disorders unless the deviance or conflict results from a dysfunction in the individual, as described above.19

Three features of this revision stand out in light of our previous discussion of the conceptual difficulties of defining mental disorder and excluding homosexuality from the definition.

First, the new definition continues to rely on ‘dysfunction’. Yet, as argued, no satisfactory account of cognitive, emotional or behavioural functions exists. The notion of dysfunction, insofar as we have briefly reviewed the most prominent definitions in psychiatry, fails to provide either a scientifically robust or a conceptually sound model. We have argued that definitions centred on ‘dysfunction’ cannot account for the exclusion of homosexuality from the category of disorder. Second, the revised definition implies that distress and impairment are not necessary conditions for a diagnosis of mental disorder, relegating their importance to that of supplementary factors (‘usually associated’) that may enable a diagnosis to be made. This goes against Spitzer's observations that evaluative criteria (including harm) ought to be central to the concept of mental disorder, which, as we have previously noted, is what swayed him towards arguing for the declassification of homosexuality.

Third, the wording of the exclusion of ‘socially deviant behaviour’ in the new definition differs from that in DSM-IV in suggesting a deviance or conflict that is the ‘result’ of a dysfunction could be a mental disorder (the previous definition stipulated that deviant behaviours are not mental disorders “unless the deviance or conflict is a symptom of a dysfunction in the individual”). The terminological difference is subtle but has implications. For example, take cognitive impairment as an example of a ‘dysfunction’. If an individual is bullied on the basis of this, then there is a sense in which conflict loosely ‘results’ from that impairment, but it is certainly not a ‘symptom’ of the impairment. However, if a particular cognitive impairment leads an individual to provoke or seek conflict to an abnormal extent, then arguably this conflict would be a ‘symptom’ (as well as a result) of the impairment. In both cases, conflicts ‘result’ from an impairment, but only in the latter case might the conflict be of direct clinical relevance to a possible diagnosis. Thus ‘symptom’ is felt to make an important distinction that ‘result’ is not able to make.

For the foregoing reasons, the DSM definition of mental disorder is not by itself sufficiently robust to preclude diagnosis of homosexuality in an oppressive society. Of course, in a non-discriminating society homosexuality would not be diagnosable on this definition, but other social contexts must be kept in mind. The first reason for this is that more oppressive contexts remain prevalent globally; the second is that more open societies continue to contain marginalised or oppressed groups whether on the basis of sexuality or other features. Third, pathologisation and misuse of psychiatry has historically been driven by and contributed to social exclusion, thus ongoing vigilance is called for: this was indeed what motivated many theorists to attempt a definition, following the declassification of homosexuality.

When homosexuality was broadly considered pathological, data citing homosexuals’ poor functioning, relationship problems and childhood traumas were based solely on clinical populations, without regard for the prevalence of homosexuality (and the lack of correlative evidence for poor functioning, trauma, abuse, etc.) in the broader population. Thus those seeking to retain homosexuality as a ‘mental disorder’ were able to cite psychodynamically described mental health problems as evidence of its pathological status. Questioning the inherent assumption that homosexuality was pathological went hand in hand with recognising that non-distressing and non-impairing homosexual traits were prevalent throughout normal populations. This was not a matter of conceptual analysis into the notion of disorder but rather a growing recognition that the assumption of pathology was mistaken: the work of Kinsey and broader changes to social values opened up possibilities for this questioning and revealed the implicit presumptions that had driven the pathologising of sexual variance.

Lessons from the controversy surrounding the declassification of homosexuality are highly pertinent to debates today over the apparent existence and prevalence of some mental disorders.

The biological or psychological bases of many conditions considered to be psychopathological have not been established, and it is not clear they will be. Such conditions are presumed pathological without adequate consideration of the prevalence of non-impairing or distressing ‘symptoms’ in the normal population. The British Psychological Society has warned that DSM-5 has not taken into account its criticisms that symptoms of personality disorders, ADHD or bipolar disorder in childhood have not been researched in the wider (normal) population: there is a prior assumption that these symptoms are pathological, and thus warrant psychiatric diagnosis, intervention and treatment.20 In the same way that the assumptions underpinning the pathologisation of homosexuality were uncovered and questioned in the past, today important and challenging questions are being asked of the status of controversial diagnoses. This is being driven by gaining an understanding of the value assumptions driving practitioners’ approaches to the conditions they deal with.


We have suggested that the exclusion of homosexuality provides a useful litmus test against which to evaluate candidate definitions of mental disorder, on account of the substantial scientific, social and political dimensions of psychiatric classification that it illuminates, in light of its declassification from the DSM in 1973. Indeed, in view of this controversy, efforts to provide a definition of mental disorder for the DSM were intended to safeguard against future abuses of psychiatry by ensuring that socially disvalued conditions could not be pathologised. However, we have argued that some of the most prominent candidate definitions in recent history fail to exclude homosexuality from the definition of disorder.

If we are correct, then the definition of mental disorder provided in DSM-5 fails to do important parts of the work that the definition, albeit with caveats, was developed to do: it fails in adjudicating disorder from non-disorder, and it fails to mitigate against potential future abuses of psychiatry. If this cannot be improved on, the evolving efforts to provide a definition of disorder appear to be increasingly dysfunctional, and future versions of the DSM face a challenge. At present, the provision of a definition provides a gloss, implying that disorders can be distinguished from non-disorders by the application of scientific-looking criteria. If this is not in fact possible, it would be better to state transparently that there is no adequate definition of disorder than reiterate the false reassurance of an impotent safeguard. While DSM-5 does acknowledge that no definition can “adequately capture all aspects of all disorders” (DSM-5, p. 20), the attempt to nonetheless provide a definition indicates that the authors still consider a definition to be doing important conceptual and practical work in distinguishing disorder from non-disorder. We suggest that in addition to the issues we have raised, there are good reasons to shift focus away from the complexities of defining mental disorder and more towards the practice of psychiatry. Understanding the assumptions that drive diagnosis and treatment by clinicians may prove more illustrative than conceptual analysis and a scientific gloss. It may after all be safer not to provide a definition at all.


Our thanks to King's College London Philosophy of Psychiatry group for discussions during the development of this paper and for their comments on an earlier draft.


View Abstract


  • Competing interests NB's contribution was supported by a Strategic Award grant from the Wellcome Trust.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • i We are working within the societal context in which homosexuality is not a disorder. This may be open to dispute in some societies, but this is a highly controversial claim and one that does not impact on our current considerations of the work done by defining mental disorder for the DSM.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles