Classifications in psychiatry: a conceptual history*

Download Classifications in psychiatry: a conceptual history*

Post on 06-Jul-2016




0 download


  • Guided by rules not always identifiable from theperspective of the present [2], most human cultures[3] have gone through periods of classificatory zealand produced more than their fair share of classesout of the objects, real and ideal, populating their uni-verse. Western culture and its medicine have not beenan exception and since Classical times, all manner ofhuman ailments, including those pertaining to behav-

    Ordinary article OA 555 EN

    Classifications in psychiatry: a conceptualhistory*

    German E. Berrios

    Background: Historical accounts of psychiatric classifications have hitherto beenwritten in terms of a received view. This contains two assumptions, that: (i) the activ-ity of classifying is inherent to the human mind; and (ii) psychiatric phenomenaa r estable natural objects. O b j e c t i v e s :The aim of this article is to provide an outline of the evolution of psy-chiatric classifications from the perspective of conceptual history. This is defined asa theoretical and empirical inquiry into the principles, sortal techniques and contextsin which alienists carried out their task. It assumes that all psychiatric classificationsare cultural products, and endeavours to answer the question of whether classifica-tory models imported from the natural sciences can be applied to man-made con-structs (such as mental illness) definitionally based on personalised semantics. M e t h o d s :Exemplars of classificatory activity are first mapped and contextualised.Then, it is suggested that in each historical period crafting classifications has beenlike playing a game of chess with each move being governed by rules. This is illus-trated by offering an analysis of the 18601861 French debate on classification. Results and Conclusions: (1) Medicine is not a contemplative but a modificatoryactivity and hence classifications are only valuable if they can release new informa-tion about the object classified. (2) It should not be inferred from the fact that psy-chiatric classifications are not working well (i.e. that they only behave as actuarialdevices) that they must be given up. Conceptual work needs to continue to identify i n v a r i a n t s (i.e. stable elements that anchor classifications to nature. (3) Becausemental disorders are more than unstable behavioural epiphenomena wrappedaround stable molecular changes, neurobiologicalinvariants may not do. Stabilitydepends upon time frames. Furthermore, it is unlikely that gene-based classifica-tions will ever be considered as classifications of mental disorders. For once, theywould have low predictive power because of their lack of information about the defin-ing codes of mental illness. Socialand psychologicalinvariants have problems oftheir own.Key words: classification, diagnosis, natural kinds, psychiatry, taxonomy.

    Australian and New Zealand Journal of Psychiatry 1999; 33:145160

    *This article is dedicated to Dr MikulsTeich, one of the great histori-ans of science of the 20th century, on his 80th birthday. I have had theprivilege of enjoying, first at Oxford and then at Cambridge, the plea-sures of his teaching and friendship.German E. Berrios, University Lecturer in Psychiatry and ConsultantNeuropsychiatrist [1]D e p a rtment of Psychiatry, University of Cambridge,Addenbrookes Hospital, Hills Road, Cambridge CB2 2QQ, UnitedKingdom. Email: Received 8 January 1999; accepted 15 January 1999.


    iour and/or the mind, have been assiduously classi-fied [4].

    One such classificatory drive, mainly concerningplants, animals and languages, appeared in the Westduring the 17th century [5]. By the end of the 18thcentury [6], it had been generalised to all Creationand crucially imbued with the belief that classify-ing was inherent to man [7], who carried it outaccording to cognitive rules written either into hisbrain [8], mind [9] or language [10]. From then on,the debate was no longer about whether one shouldclassify, but how [11]. It has remained so to this day.According to the majority, groupings have to beformed in terms of privilegedfeatures [12] (e.g. thereproductive organs of plants). Linn and other 18thcentury botanists applied this method to diseases[13]. Towards the end of the 18th century, Adanson[14] dared to suggest that all features should be takeninto account; his insight is now considered as a pre-cursor of numerical taxonomy [15].

    It is currently accepted that all classificatory actsmust assume a theory of concepts[16]. Eighteenthcentury classificators did likewise sharing the beliefthat concepts were sortalinstruments (i.e. devicesto create classes by comparing candidate items witha paragon). The latter, in turn, was conceived aseither a list of features (menu approach) or an image(a prototype) with sufficient discriminatory power tosort out all denizens of a finite universe into membersand non-members of the given class [17]. The nature,origins and construction rules for the paragon wereno longer discussed, for, by then, Lockes account ofthe meaning of terms (central to his theory of class-ification) had been widely accepted [18].

    A second important 18th century assumption wasthat the naturalclassificatory drive exhibited byman was innate (i.e. God-given). However, as evolu-tionary principles became established during the 19thcentury, the competing belief emerged that classify-ing was an adaptativefunction (i.e. that it had beenselected inby nature for it conferred a cognitiveadvantage). All of these earlier assumptions are stillembedded in current classificatory ideas, includingthose reigning in psychiatry.

    In suchZeitgeist, classifying mental illness (or itsphenomena) is mandatory, and as it was in 1798, in1998 the mandate remains unchallenged. In roughand ready summary, views since the 19th century(which is the historical period to be dealt with in thispaper) can be grouped according to whether theyassume that the principles of scientificc l a s s-ification are universal and: (i) apply to psychiatry

    tout court [19]; (ii) apply to psychiatry mutatismutandi[20]; or (iii) they do not apply at all, for sodoing would be an instance of ignoratio elenchi[21,22].

    To the question, how it is to be decided which ofthese three views is the correct one, the conventionalanswer has been by interrogating nature(i.e. byempirical research). However, this answer makestwo unwarranted assumptions: that nature is a trea-sure of ready-made knowledge[23], and thatempirical research is theory-free. It is more likely,however, that at the moment the question has noanswer for, so far, no science (including physics) hasbeen able to develop a crucial empirical testthatcan decide between rival conceptual systems. Thereis the additional complication that empiricalresearchis wedded to a view of nature according towhich nature and her objects are fully independentfrom the mind, have a mathematical structure, etc.,and hence it cannot be used to compare two differentepistemes, particularly when one of them may actu-ally reject that particular view of nature. In view ofall this, to expect psychiatry to develop a crucial testis utopian (i.e. by empirical research alone it will notbe able to identify, now or ever, a realclassificationof mental disorders) [24].

    A more realistic solution might, therefore, be forempirical work only to start once an agreement orconvention has been reached between the variousparties on matters such as: (i) the nature of the phe-nomena to be classified; (ii) the type of conceptandclassto be used; and (iii) a sortingmethodology.Because such agreement also requires that the con-ceptual analysis [25] be focused on a specific timeand space frame (i.e. region of the world), it becomesobvious that a conceptual historyof psychiatricclassifications is needed on which to base the agree-ment [26]. As far as I know, the latter has not yetbeen written. It is not that writings on psychiatricclassificationsor their history do not abound, forsince the 19th century the literature in this field hasgrown at much pace. It is that this material is sur-prisingly repetitious and parochial, and oftenintended to serve too many masters (e.g. clinical,administrative, didactic and propagandistic).Furthermore, they are based on inappropriate histori-ographical techniques and are not able to tell us aboutthe history of concepts.

    Nonetheless, these writings offer rich pickings. Inaddition to containing factsabout earlier classifica-tions (in the manner of good catalogues), they tellsomething about the scientific and social world in

  • G.E. BERRIOS 147

    which classifications were born. For example, duringthe 19th century, it was part of professional growthand success that alienists developed a personalclassification [27]. These writings also tell about thedeeper forces (social and otherwise) that have alwayscontrolled the methodology and implementation ofclassifications. Depending upon faon de parler, his-torians may say that such forces originate from aparadigm, cognitive style, epistemeor fashion.The advantage of the last term is that it, in oppositionto the others, refers to a resistible social process, andhence removes the alibi that all alienists were con-trolled by ineluctable cognitive rules[28].

    By providing an outline for a future history of psy-chiatric classifications, this paper aims at helpingreaders escape from the conceptual myth that psychi-atrists are obliged to classify mental disordersbecause that is the way in which the brain or lan-guage works. Because of its non-specificity (i.e. itcan be marshalled in favour of any cause), this typeof argument is unhelpful. Nonetheless, it is fair to saythat classifications may serve a number of house-keeping and actuarial functions (e.g. providing anomenclature, furnishing a basis for informationretrieval, and description) [29]. In view of what willbe discussed in this paper, it is less clear that they canalso help with prediction and theory formulation.Little will be said here about the sortal act itself (forone, this raises the ugly problem of symptom anddisease recognition) [30] or about the current prod-ucts of classification.

    Definitions now and in history

    Classificationis an ambiguous term in that it mayrefer to: (i) the creation of principles and rules interms of which items, referents, features, and futuregroupings are to be constituted (what is called taxon-omy when related to animals, plants and otherobjects of the tangible world) [31]; (ii) the act ofgrouping itself (sortal act; what in the case of psychiatry partially corresponds to diagnosis); and(iii) the products of activities (i) and/or (ii) (ICD-10and other classificatory systems can be considered asan exemplar of this). There is no space in this papereven to start discussing the implications and func-tionalities attached to each of these meanings.

    Let us start with a 20th century definition:Classification is the identification of the category orgroup to which an individual or object belongs on thebasis of its observed characteristics. When the char-acteristics are a number of numerical measurements,

    the assignment to groups is called by some statisti-cians discrimination, and the combination of mea-surements used is called a discriminant function[32].

    There is little difference between the above and a19th century definition: In every act of class-ification, two steps must be taken; certain marks areto be selected, the possession of which is to be thetitle to admission into classes, and then all the objectsthat possess them are to be ascertained. When themarks selected are really important and connectedclosely with the nature and functions of the thing, theclassification is said to be natural; where they aresuch as do not affect the nature of the objects materi-ally, and belong in common to things the most differ-ent in their main properties, it is artificial ... theadvantages of classification are to give a convenientform to our acquirements, and to enlarge our knowl-edge of the relations in which different objects standto one another...[33].

    There now follows an 18th century view: All[beings] are nuanced and change by degrees in nature.There is no one being that does not have either aboveor below it another that shares with him some featuresand that differs from him in other features[34]; and a17th century one: Nature, in the production ofdisease, is uniform and consistent; so much so, that forthe same disease in different persons the symptoms arefor the most part the same; and the selfsame phenom-ena that you would observe in the sickness of aSocrates you would observe in the sickness of a sim-pleton. Just so the universal characters of a plant areextended to every individual of the species; andwhoever (I speak in the way of illustration) shouldaccurately describe the colour, the taste, the smell, thefigure, &c. of one single violet, would find that hisdescription held good, there or thereabouts, for all theviolets of that particular species upon the face of thee a r t h [35]. The common points shown by these defin-itions reflect their common origin and hence thesharing of assumptions about both the entities to beclassified and the rules for their classification.


    Any metalanguage (i.e. a second-order language)must include a minimum set of conceptual tools toallow the examination of the first-order languageitself, in this case taxonomy (as opposed to theobjects in the world to which the theory applies). Ametalanguage for classificatory theory is graduallybeing constituted. The first distinction to be made


    concerns the classificatory system and its domain:the former refers to a frame, the set of rules and thedefinition of the basic units of analysis; the domainconcerns the population or universe of entities to beclassified. These elements interact with one anotherbut of particular relevance is the interaction betweenthe definition of the units of analysis and its domain.This because many might say that the former helps toconstitute the latter.

    The metalanguage of classification has also beenexpressed in terms of a number of dichotomies: cat-egorical versus dimensional; monothetic versuspolythetic; natural versus artificial, top-to-bottomversus bottom-up, structured versus listing, hier-archical versus non-hierarchical, exhaustive versuspartial, idiographic versus nomothetic, etc. Thesepairs of concepts criss-cross one another and it is noteasy to deal with them independently. Some concernfeatures or dimensions, others the compass of classi-fications, yet others relate to the methodology of tax-onomy. Hence, only some may apply to psychiatry.

    Ambiguity abounds within dichotomies. Forexample, in the dichotomy categorical versusdimensional, the first term can be interpreted asreferring to the status of features or properties (afeature is either present or absent or 0 or 1) or to thestyle of sorting (i.e. an entity is or is not a member ofa class); likewise the second term has a two-foldinterpretation:...