Psychological Medicinehttp://journals.cambridge.org/PSM

Additional services for Psychological Medicine:

Email alerts: Click hereSubscriptions: Click hereCommercial reprints: Click hereTerms of use : Click here

The placebo: from specificity to the non­specific and back

Michael Shepherd

Psychological Medicine / Volume 23 / Issue 03 / August 1993, pp 569 ­ 578DOI: 10.1017/S0033291700025356, Published online: 09 July 2009

Link to this article: http://journals.cambridge.org/abstract_S0033291700025356

How to cite this article:Michael Shepherd (1993). The placebo: from specificity to the non­specific and back. Psychological Medicine, 23, pp 569­578 doi:10.1017/S0033291700025356

Request Permissions : Click here

Downloaded from http://journals.cambridge.org/PSM, IP address: 147.143.4.20 on 26 Feb 2013

Psychological Medicine, 1993, 23, 569-578. Copyright © 1993 Cambridge University Press

The placebo: from specificity to the non-specificand back1

MICHAEL SHEPHERD2

From the Institute of Psychiatry, London

SYNOPSIS A brief historical outline is provided of the concept of specificity in biology andmedicine. The advent of scientifically based therapeutics, and especially clinical pharmacology, hasdirected increasing attention to the role of non-specific factors in the field of mental disorders. Theactions and effects of placebos are discussed in relation to the specific/non-specific dichotomy withparticular reference to both pharmacological and psychological modes of treatment.

INTRODUCTION

'The properties of placebos are said to be non-specific in that they are largely independent ofthe chemical substances of which they arecomposed and in that they are manifest in sowide a variety of disorders as to precludeexplanation by a single mechanism' (Eisenberg,1984). This serviceable definition, like mostothers, can be criticized according to thestandpoint adopted but in common with themajority it includes a reference to the so-called'non-specific' factors with which the placebohas been generally associated. The large litera-ture on the concept and semantics of 'non-specificity', however, rarely takes cognizance ofthe still larger literature on specificity, principallybecause this has been primarily the concern ofhistorians and practitioners of biological scienceand bio-medicine. A glance at the compendiousvoume Placebo: Theory, Research and Mec-hanisms, edited by White, Thursky and Schwartz(1985) and dominated by psychologists, illu-strates the point. If, as I would maintain, it islogical to approach the matter by proceedingfrom the positive to the negative, priority shouldbe given to specificity in its own right beforeproceeding to the domains of non-specificitywithin which the placebo concept is to be located.

1 Based on a lecture delivered at the 50th Anniversary InternationalMeeting of the American Psychosomatic Society in New York City,1992.

! Address for correspondence: Professor Michael Shepherd,Institute of Psychiatry, De Crespigny Park, London SE5 8AF.

In this context a brief historical outline of theirdevelopment, primarily from the standpoint ofclinical science, provides a background tocurrent empirical and conceptual enquiry.

BIOLOGICAL SPECIFICITY

The notion of specificity has been a constantpre-occupation among physicians since the timeof ancient Greece. Essentially, as Owsei Temkinhas pointed out, it is bound up with a funda-mental dichotomy contrasting the Platonic or'ontological' with the Hippocratic or 'physio-logical' view of disease (Temkin, 1963). Theontological notion of disease postulates anindependent, self-sufficient entity with its ownnatural history; the 'physiological' emphasizesthe individual biography of the patient. Thedistinction reflects the familiar contrast drawnbetween disease in man and man in disease.

Over the centuries the pendulum of opinionhas oscillated between these two standpoints.Tracing its evolution, Galdston has observedthat after centuries dominated by the thinking ofHippocrates and Galen the ontological conceptof specificity came back into its own with theRenaissance, first through Paracelsus and thenthrough the work of such men as WilliamHarvey, Robert Boyle, Francois Boissier deSauvages and, later, Morgagni, Bichat andRokitansky (Galdston, 1941-2). All of themaccepted the Cartesian model of the humanbody as a machine and, as Galdston remarks:'Always we find that when the mechanistic

569

570 M. Shepherd

viewpoint prevails in the biological sciences,medicine is specifistic in its nosology, aetiologyand therapy'.

During the nineteenth century the physio-logical school produced a doughty champion inthe person of Rudolf Virchow. In his forthrightpolemic, Specifiker und Specifisches, Virchow(1854) stated the case with characteristic di-rectness. 'The destruction of the ontologicalconception of disease leads also to the de-struction of ontological therapy, the School ofthe Specifists. Therapy is confronted not bydiseases but by conditions; everywhere we dealonly with alterations in the "life circum-stances'". By the end of the century, however,the specifists were back in the saddle with thearrival of the bacteriological era. Only slowlyhas the pendulum swung back during thetwentieth century. In Britain the challenge wasmounted in the name of social medicine (Ryle,1944). In American medicine Richard Shryock(1969) has written perceptively on 'The Conceptof the Specific', alluding to the influence of thestress concept with which psychosomatic medi-cine was so closely linked in its early days. Atthat time psychosomatics had to contend withits own 'specificity' theory when Franz Alex-ander, using a psychoanalytical framework,postulated specific unconscious and unresolvedconflicts as being responsible for particularsomatic diseases (Alexander & Szasz, 1952).These conflicts supposedly determined thenature of the emotion that is suppressed orunconscious, resulting in a state of chronicity.This hypothesis of a 'specific emotion' wasrelated to Flanders Dunbars' concept of aspecific personality and to Graham's, notion ofspecific attitudes. Empirical research subse-quently failed to support these notions, andStewart Wolf (1961) summed up the situationsome years ago in his paper' Disease as a way oflife: neural integration in systemic pathology',suggesting that 'physicians now consider mostdiseases to be distinct from one another insofaras they represent patterned responses or adapta-tions to noxious forces in the environment'. Thecurrently fashionable biopsychosocial model ofillness turns on the same paradigm.

This long-standing debate within medicinehas been reflected by a corresponding trendwithin biological science since its inception.Indeed, Lester King (1978) has gone so far as to

argue that the history of physiology is syn-onymous with the search for specificity. Cellularspecificity has been a basic concern of embryo-logists since the late nineteenth century, andJoseph Needham has drawn a direct parallelbetween medical and the biological thinking:

Is not the ontological-physiological antithesis in thephilosophy of pathology a special case of the fullygeneralized biological situation of special stimulus onthe one hand, and cell reactivity on the other?... Afterthe classical discovery of the embryonic organiser-region by Hans Spemann in the twenties, and the noless classical discovery ten years later that the activeprinciple of the gastrula's dorsal blastopore lip wasstable to boiling, a stream of research started which isyet far from exhausted. Some (including myself) havealways been convinced that in normal development aspecific chemical substance, a Wirkstoff, is primarilyinvolved, but many biologists have preferred... toplace most of the emphasis upon the reactivity of theoverlying ectoderm, hence on the cell-proteins of thisgerm-layer, which must convert it into neural tissue asdevelopment goes on. In the end, Hippocrates mustalways be right - we cannot do without both of thefactors, external and internal; the only problem is toknow how they interact (Needham, 1963).

Nonetheless, the term 'specific' is so widelyused in biological research that Peter and JeanMedawar have commented that:

If an estimate were to be made of the technical termsmost frequently used in the professional writing anddialogue of biologists, we have little doubt that'specificity' would come out way ahead. Specificityrefers to the complementarity or matched opposite-ness in reactions between antigen and antibody,enzyme and substrate, and between stimulus andresponse in reflex action (Medawar & Medawar,1985).

To underline the point mention may be made ofthree outstanding contributions that illustratethe variety of usages employed by biologicalscientists. First, Linus Pauling's fundamentalwork on molecular biology, which was foundedon complementary action (Pauling, 1940). Sec-ondly, Roger Sperry's work on the synapse inthe visual system, which was directed at under-pinning the specificity of cellular function withinthe central nervous system (Attardi & Sperry,1963). Thirdly, R. A. Fisher's 'specific modu-lator', which played a large part in formulatingmodern notions of gene function (Fisher, 1930).

The placebo: from specificity to the non-specific and back 571

PHARMACOLOGICAL SPECIFICITY

The biological domain with which we are mostconcerned, however, has to do with pharma-cology because of its direct links with thera-peutics. For proponents of the ontological viewof disease the search for specific remedies hasbeen a constant goal. The early work oncolchicine or cathartics is primitive compared tomodern chemotherapy, but the aims are identical(Temkin, 1972). In his monograph entitled Ofthe Reconcileableness of Specifick Medicines tothe Corpuscular Philosophy, published in 1685,Boyle even anticipated the standpoint of Ehrlichand the era of antibiotics (Boyle, 1685). Only inthe recent past, however, has it been possible toidentify some of the mechanisms involved. Inthe 1930s Clark proposed a general theory ofantibiotic drug-mechanisms in terms of toxicaction on nuclear processes (Clark, 1937),observing at the same time that whereas somesubstances, like colchicine, induced a selectivetoxic effect others, like phenol or formalin,induced what he called non-specific toxic effectsby deranging the cellular protein structure.External influences like heat, oxidation orirradiation were also, he observed, best regardedas non-specific toxic agents acting on enzymes.Clark also extended his concept of the specificityof drug-action to the sphere of drug-antagonism.Here he distinguished chemical and physio-logical from specific forms of antagonism,reserving the latter term for cases in which onedrug inhibits the action of another drug onliving cells, though there is no drug-interactionin vitro.

10-8 10-7 10"6

Displacer (mol/1)

FIG. 1. Basic principles of receptor binding. (From Muller, 1987.)

With the identification of neurotransmitterscame a more sophisticated view of specificity,derived originally from the actions of acetyl-choline and well summarized by Sir John Eccles:

we assume that the post-synaptic membrane of thesynapse has a highly developed specificity to thesynaptic transmitter substance and its analogues. Therelationship resembles that of lock and key. One ofthe major problems confronting neuropharmacologyis to find the chemical keys for the synaptic locks. Themore specific the keys, the more discriminate will bethe pharmacological actions. However, on the otherhand, there will also be value in having 'skeletonkeys' that open closely related types of lock (Eccles,1959).

Paradoxically, this pharmacological questfor the specific has served to highlight the bio-logically non-specific, nowhere more strikinglyso than in the study of the benzodiazepinereceptor, the highly specific benzodiazepinebinding site in the central nervous system.Originally the receptor-mediated effects of drugswere examined indirectly by the use of specific orcompetitive antagonists. This method has beenemployed with benzodiazepine receptor antag-onists like imidiazabenzodiazepinones, beta-carbolines and phenylpyrazoloquinolines, butwith the advent of direct binding techniquestritiated benzodiazepines were used as radio-ligands. There is now extensive evidence sup-porting the view that the benzodiazepines act asGABA-mimetic substances and that receptoraffinity represents the biochemical correlate ofthe therapeutic dose in man, GABA itself beingan important inhibitory neurotransmitter of themammalian central nervous system, while thebenzodiazepine receptor is widely regarded as amodulatory unit of GABA-ergic neurotrans-mission. Yet, as Muller points out:

Unfortunately, practically all radio-ligands bind notonly to their receptors, but also to different degrees tomuch less specific binding sites on the tissuesinvestigated or even on the filters used for separatingbound from free ligand. This unspecific or non-specific binding varies with the ligands and tissuesused can account for less than 5 % up to more than90 % of total binding... receptor binding (also referredto as 'specific binding') and non-specific bindingdiffer in some major aspects (Muller, 1987). (Fig. 1).

Even specific binding, it appears, is affected bymany factors, e.g. ligand concentration, cerebralregion, age, dosage, route of administration,

572 M. Shepherd

and the presence of other drugs (Garattini &Mennini, 1986). Further, there appear to be atleast two subclasses of receptor, one of whichbinds ligands with more affinity than the other,and it still remains uncertain whether thebenzodiazepine receptor has an independentphysiological function and whether an endogen-ous ligand is involved. In addition, peripheralbenzodiazepine binding sites have been estab-lished which are specific in respect of theirbinding capacity but are pharmacologicallyinactive.

NON-SPECIFICITY AND THE PLACEBO

From this and comparable work with otherneurotransmitters biological specificity emergesmore as a metaphysical than a physical concept,to be viewed in 'more or less' rather than' absolute' terms. The consequence most relevantto our present concerns has been drawn by S.Fisher:

There is no clear-cut agreement as to what, specifically,is meant by the phrase 'specificity of drug action'.And since the definition of 'specificity' is so vague,one cannot expect too much from its 'non-specific'corollary (Fisher, 1970).

The validity of this conclusion, based as it ison laboratory data, is greatly extended by thereal-life human transaction of medical care inwhich a host of non-biological factors must betaken into account. Writing as a self-styled' sociopsychopharmacologist', Joyce has listedsome of the more important (Joyce, 1989)(Fig. 2).

Several of these psycho-social factors may, ofcourse, be specific in their own right. The clinicalnotion of 'reactivity' implies the possibility ofnormalizing a morbid state if the noxa can beidentified and reversed - as, for example, withhospitalism, transient personal loss or economicmisfortune - and Harris and Brown have com-mented on the specificity of some life-events inprovoking particular types of disorder (Harris &Brown, 1989). Nonetheless, the tendency togroup most psycho-social factors together asnon-specific has diminished scientific enquiryinto their mode of action until recently, a trendthat has been stimulated by the increasingattention paid to therapeutic specificity. Among

the fruits of this concern has been the notion ofthe placebo, literally 'I shall please', reifying averb into a noun and introducing the element ofsubjectivity into the operational matrix.

In her brief historical survey Suzanne White(1985) has claimed that the first acknowledge-ment of the status of the placebo, which she calls'medicine's humble humbug' is to be found inOliver Pepper's 'A note on the placebo' pub-lished in the American Journal of Pharmacologyin 1945 (Pepper, 1945) and pin-pointing theplacebo as a biologically inactive form ofmedication. According to Pepper, its prescrip-tion demands 'a certain amount of skill'; itshould have 'a Latin and polysyllabic name'; itshould be given with 'some assurance andemphasis for psychotherapeutic effect'; and,above all, it must be seen a s ' a part of what usedto be called the art of medicine - the discreditedsister of scientific medicine'. Since then theplacebo has attracted the attention of workersfrom four disciplines in turn, namely phar-macology, psychology, philosophy and bio-chemistry. Their differences in outlook partlyexplain the confusion attending the semantics ofthe placebo which have been extensively dis-cussed by Shapiro (1968), Brody (1977) andGninbaum (1981).

Chronologically, the conferment of scientificrespectability on the placebo in the post-warperiod was initiated principally by workers inthe new field of clinical pharmacology. Firstcame the studies of Beecher who demonstratedthat the placebo-effect was therapeuticallypowerful and could be demonstrated in manymorbid conditions (Beecher, 1955). Why thiswas so, however, remained obscure. Reviewingthe subject in Scientific American in 1955Lasagna remarked that '...most of these "prin-ciples" of the art of the placebo are based not onany systematic investigation of the facts but onimpressions. Almost no controlled studies ofthem have been made' (Lasagna, 1955). Shortlyafterwards he entered a plea for what he calledthe' sociopsychology of therapeutics', comment-ing that

therapeutics has paid only lip service for so long tothe importance of psychological and sociologicalvariables. Too often we have had a sterile dichotomybetween those individuals who tend to forget var-iability of response and the importance of non-drugvariables, and those individuals so obsessed with the

The placebo: from specificity to the non-specific and back 573

A partial listing of non-specific influences of treatment.

Environment (milieu exterieur)-Animate

-Medical-Family-Friends-Casual contacts

-Inanimate-Physical

-Temperature, etc.-Light/dark

-Chemical-Pollutants-Diet

Patient (milieu interieur)-Physiological

-Metabolism-Genetics-Structure-Aging

-Psychological-Experience-Expectations-Personality

Drug-Size-Shape-Color-Taste-Route

FIG. 2. A partial listing of non-specific influences of treatment.(From Joyce, 1989.)

complexity of variables involved in drug responsesthat they are intellectually paralyzed (Lasagna, 1962).

Secondly, came the entry of the placebo into thefield of experimental medicine via clinical thera-peutics. Sir George Pickering asserted in 1949,

Therapeutics is the branch of medicine that, by itsvery nature, should be experimental... Before con-cluding that the change for better or for worse in thepatient is due to the specific treatment employed, wemust ascertain whether the result can be repeated asignificant number of times in similar patients,whether the result was merely due to the naturalhistory of the disease or in other words to the lapse oftime, or whether it was due to some other factorwhich was necessarily associated with the therapeuticmeasure in question (Pickering, 1949).

This passage is quoted with approval by SirAustin Bradford Hill, the founder of the modernclinical trial, which has become one of themainstays of modern medicine and which makesextensive use of the placebo or' dummy'. Writingof the use of placebo treatment he says:

The object is twofold. On the one hand we hope tobe able to discount any bias in patient or doctor intheir judgements of the treatment under study. For

example, a new treatment is often given morefavourable judgement that its value actually warrants.The effects of suggestibility, anticipation and so onmust thus be allowed for. In addition, the placeboprovides a vital control for the frequency of spon-taneous changes that may take place in the course ofa disease and are independent of the treatment understudy.

In these two ways the placebo aids us to distinguishbetween (a) the pharmacological effect of a drug and(b) the psychological effects of treatment and thefortuitous changes that can take place in the course oftime (Hill, 1971).

The point is illustrated by Fig. 3.

141 *T

13121110

o 9£ 8W 7

654321

1

3

//

//

NDrug effect

• • . , ^

'Placebo effect'

— Active drug

- No treatment

\\

\\

\\

— - - •

0 1 52 3 4

FIG. 3. Placebo and drug effects over time. (From Joyce, 1989.)

Accordingly, the central task of the phar-macologist becomes the search for specificityand, by the same token, the elimination of the'non-specific'. This view has been the corner-stone of the various forms of clinical trial, theprimary function of the placebo being toeliminate bias. However, valuable as this methodof investigation has proved to be, it is not alwaysrecalled that the controlled trial is essentially anexercise in experimental epidemiology and assuch has to do primarily with populations andhence with averages rather than with individuals.This point was made with some force by LancelotHogben (1954) for whom the aim of theassessment of remedies entailed ' the disclosureof a specific stimulus-response nexus', somethingthat can never be attained by the study ofgroups. For the most part the data derived fromclinical trials are stochastic. The establishmentof biological specificity is therefore unattainableby the standard techniques associated with thismethod of evaluation. Hogben's own attempt to

574 M. Shepherd

achieve this goal within a therapeutic frameworkwas via the self-controlled trial on a singleindividual but this, as he conceded, is limited tolow-grade, reversible forms of morbidity (Hog-ben & Sim, 1953).

THE PLACEBO EFFECT ANDPSYCHOTROPIC DRUGS

Despite the manifest importance of the placeboeffects a glance at most textbooks of clinicalpharmacology shows that they were, and tosome extent still are, largely ignored or mini-mized in discussing the treatment of physicalillness. In part this reflects the emphasis ondisorders with well-established aetiology anddemonstrable response to rational treatment.Brody has observed 'that one of the moststriking features of the placebo phenomenon isits ability to effect change in virtually anypotentially reversible disease or disorder, whilecontemporary theories of pathophysiology andtherapeutics lead us to expect a "specific" orcharacteristic remedy to be efficacious for only asmall number of disorders. We anticipate that a" specific " / " characteristic " remedy has roughlya 75-95 % probability of being efficacious for avery small number of disorders, while a placebohas a 30-40 % probability of being efficaciousfor almost any disorder' (Brody, 1985). Thisbeing the case, it was to be expected that thepharmacotherapy of psychiatric and 'psycho-somatic ' disorders would lend itself to a moreproductive study of the placebo. In this con-nection it may be recalled that the pioneeringinvestigations of Emil Kraepelin in WilhelmWundt's psychological laboratory in the 1890shad resulted in studies of what he calledpharmacopsychology and had led him toconclude that 'we might be able to learnfrom a specific drug or a specific symptom some-thing about the true nature of this symptom'(Kraepelin, 1892).

The arrival of the new psychotropic drugs inthe 1950s and the subsequent formation ofpsychopharmacology as an independentdiscipline enlarged the spectrum of phenomenaunder investigation by incorporating traditionalpsychological topics within the framework ofenquiry. In particular, the newer techniques ofbehavioural pharmacology brought experimen-tal psychology into play and the ' specific' -

' non-specific' dichotomy was given a new formatin terms of reflex responses. Cook, for example,stated in 1959:

Animals which had their conditioned responseblocked and still maintained unconditioned responses,were considered to have exhibited a specific block ofthe conditioned responses. Animals which had bothconditioned and unconditioned responses blockedwere considered to have exhibited a non-specificblock (Cook, 1959).

The flush of therapeutic enthusiasm engen-dered by the new psychotropic drugs, however,tended to overshadow the attention drawn byCheney & Drewry as early as 1938 to what theycalled the 'non-specific' factors influencingthe response to deep insulin treatment ofschizophrenia (Cheney & Drewry, 1938). Theseprimarily psycho-social factors included staffattitudes, patient expectations, improved wardmilieu, and changing social policies.

A striking example of this process wasprovided in the mid 1950s by the introduction ofthe first wave of psychotropic drugs. Whilepharmacologists speculated on their mode ofaction and clinicians on their modification ofpsychopathology the major impact was on thepopulations of mental hospitals. The movementtowards ' deinstitutionalization' was initiatedby claims that drug-treatment was making itpossible to open the asylum-doors, nowhereadvanced more firmly than in New York Stateby Brill and Patton 1955). In Britain there was asharp public disagreement between two eminentleaders of the psychiatric fraternity. For one, aconvinced somaticist, 'hospital administrativereforms would be crying in the wilderness if theso-called tranquillizing drugs had not appeared'(Mayer-Gross, 1959). For the other, representinga much more broadly based view, 'if we had tochoose between abandoning the industrial re-habilitation centres and other social facilitiesavailable to us there would be no hesitationabout the choice - the drugs would go' (Lewis,1959).

How can a major issue of this kind beresolved? Most of the clinical trials were on toosmall scale to be conclusive, but we were able tocarry out a detailed trend-analysis of mentalhospital population statistics before and afterthe introduction of the psychotropic drugs(Shepherd et al. 1961). The hospital in question

The placebo: from specificity to the non-specific and back 575

had adopted a forward-looking policy longbefore the new drugs were available, and thedata showed that the impact of medication onthe movement of in-patients was relatively small.

This result was in line with 0degaard's studyof the variation between the hospital statistics ofNorwegian institutions, demonstrating para-doxically that 'in hospitals with a favourabletherapeutic situation the psychotropic drugsbrought little, or no improvement in the rate ofdischarges. In hospitals with a low pre-drugdischarge-rate, on the other hand, the improve-ment was considerable (0degaard, 1964)'. Inother words, the non-specific placebo and milieueffects become most evident when the therapeuticsituation is least favourable.

In the light of such findings it becomes patentlyinadequate to identify and evaluate non-specifi-city as no more than the residual obverse ofspecificity. The logic of this viewpoint has beencogently developed by Paul Meehl who analysedthe part played by what he calls 'stronginfluences' which are not specifically causal(Meehl, 1977). Drawing on J. L. Mackie's ac-count of causality, Meehl suggests that themodal type of causal explanation in the medicaland social sciences is not biologically specificand is, indeed, fundamentally non-specific. Fur-ther, the use of the word 'cause' is legitimateeven though the attributed cause is neithernecessary nor sufficient in Koch's sense of theterms. What is sufficient is the entire complex ofcircumstances, and he summarizes the positionin the following sentence:' A causal factor c maybe an insufficient but necessary part of a causalcomplex C, which complex is an unnecessarybut sufficient condition for producing the effect'.Meehl calls this the INUS condition, theacronym INUS being formed by the initialletters of I (insufficient), N (Necessary), U(unnecessary) and S (sufficient) in the sentencegiven above. This view, it may be observed,highlights the importance of probabilistic andmultiple determinants in non-specific aetiology.Applying this mode of reasoning to the factorsbearing on deinstitutionalization, the drugs maybe seen as constituting an insufficient butnecessary part of just one causal complex thatwas responsible for the massive programme ofpatient discharge. This particular complex issufficient for the purpose, but that it is un-necessary is shown by the fact that another,

drug-free complex of circumstances can producethe same effect.

THE PLACEBO AND PSYCHOTHERAPY

In this case the components of Meehl's INUS,his 'complex of circumstances', would presum-ably include a variety of psychosocial factorsfavouring discharge from hospital. Many at-tempts to account for such non-specific pheno-mena have been couched in terms of descriptiveor 'folk' psychological concepts like 'faith','expectation', 'hope', 'anticipation' and 'sug-gestibility'. All healing, as Thomas Sebeokremarks, is accompanied by the healer's ' verbalpatter' and 'strings of non-verbal signs' whichhave been recognized, often implicitly, as potentsince the dawn of medicine (Sebeok, 1983).Understandably, therefore, such phenomenahave attracted the interest of workers attemptingto unravel the nature of psychotherapy becausethey contain so many of the elements associatedwith psychotherapy itself. As Downing andRickels maintain, in the sphere of mentaldisorders the more intense the doctor-patientrelationship, the less likely is one to obtain asignificant difference between drug and placebo(Downing & Rickels, 1978). In addition, thedoctor-patient nexus is readily available forexperimental manipulation.

Empirically, one of the most thoroughgoingattempts to investigate the issues to date is thelarge-scale NIMH Treatment of DepressionCollaborative Research Program (TDSCRP),whose major purpose has been to assess theeffectiveness of two forms of brief psycho-therapy, compared with a reference form ofpharmacotherapy and a placebo for a definedgroup of depressed out-patients (Elkin et al.1988a). The study raises a number of conceptualand methodological issues which the authorshave explored in some detail. As they comment:

A comparative study of psychotherapy and pharma-cotherapy may be conceived as a 'horse race' todetermine which of the treatments is more effectiveor, alternatively, as an attempt to address moredifferentiated and ultimately more meaningful, ques-tions regarding which patients may be helped more byeach of these approaches and in what ways (Elkin etal. 19886).

In their view answers to these questions turnon the assumption that the basic differences

576 M. Shepherd

between the treatments reside in the contrastbetween their active ingredients and in themechanisms by which they effect change. Anti-depressant drug treatment is assumed to operatemainly through biological mechanisms, psycho-therapy through psychological mechanisms thatare initiated by various aspects of the interactionbetween therapist and patient. However, as theyadmit:

a design including both a pill placebo and apsychotherapy placebo is not easy to execute, largelybecause of the difficulty of devising an adequatepsychotherapy control condition. A major methodo-logical advantage of psychopharmacological outcomeresearch has been the availability of inert pill placebosthat produce no physiological effects related to thetreatment of the disorder and can thus control for thenon-active components in the drug condition. Thetask of developing a ' placebo' condition for psycho-therapy is far more complex, largely because of theproblems in articulating what the active ingredientsactually are in any form of psychotherapy (Elkin et al.1988 c).

To circumvent this obstacle the authors aimedat a less ambitious goal, preferring to studyparticular factors that may exercise an effect onoutcome, either independently or through in-teraction with a specific treatment approach. Inso doing they admit to entering the area of' non-specific' factors which operate in every form oftherapeutic intervention. They conclude that inpsychotherapy any comparison is between notjust specific techniques of intervention butbetween 'different treatment packages' each ofwhich is made up of a particular treatmentapproach and a particular therapist and inwhich the patient's expectations and preferencesplay a significant role. Shades of Meehl's INUS!

This view runs counter to the many evaluativestudies of psychotherapy that have been basedon the assumption that it is sufficiently specificto be compared with placebo controls of thetype included by Prioleau et al. (1983) in theirmeta-analytical study, e.g. pills, play-reading,group activities, film-watching, relaxation train-ing, music, personal hygiene, etc. It is more inline with the second of Jerome Frank's postu-lated two components of all treatments, the firstaimed at correcting the specific pathologicalprocesses underlying a particular morbid state,and the second at counteracting the non-specificdemoralizing impact of all illnesses (Frank,

1989). Both placebo and psychotherapy counter-act demoralization, he argues, by a combinationof 4 circumstances: (1) an emotionally charged,confiding relationship with a helping person; (2)a healing setting; (3) a rational, conceptualscheme or myth; and (4) a ritual. On this basishe concludes that as a symbolic communicationthat combats demoralization by inspiring thepatient's hopes for relief, administration of aplacebo is a form of psychotherapy (Frank,1983).

MEANINGS AND MECHANISMS

This whole process reflects what Frank calls' thetransformation of meanings' (Frank, 1986), aphrase that echoes Adolf Meyer's 'symbol-ization' as a basic function of human con-sciousness. It also introduces a hermeneuticdimension into the placebo-effect which is to bedetected among the opinions of other workers.According to Howard Brody's' meaning model',for example, a positive placebo effect is mostlikely when (a) the meaning of the illness isaltered in a positive manner, given the patient'spre-existing belief system and world view, (b) thepatient is supported by a caring group, and (c)the patient's sense of mastery and control overthe illness is restored or enhanced (Brody &Waters, 1980). Kleinman's anthropological ap-proach brings 'meaning' into line with 'expec-tation' and the 'cultural dimension' (Hahn &Kleinman, 1983), and Jerome Bruner has madea strong case for re-introducing what he calls'meaning-making' as a core activity in modernpsychology: '...culture and the quest for mean-ing within culture', he claims, 'are the propercause of human action. The biological substrate,the so-called universals of human nature, is nota cause of action but, at most, a constraint uponit or a condition for it' (Bruner, 1990).

Overtones of the old body-mind issue aresounded by this extension of the specific-non-specific dichotomy. Clarification may event-ually be obtained, at least in part, by researchinto the mechanisms of the placebo effect. AsFisher observed 20 years ago:

the phrase 'non-specific factors' is frequently usedalmost as a wastebasket category. We search forobvious physiologic or pharmacologic variables toaccount for our observed differences and if we cannotfind satisfactory explanations within this domain, we

The placebo: from specificity to the non-specific and back 577

then invoke the concept of'non-specific', 'non-drug'causal factors...many of the non-drug factors mayeventually turn out to be typical conventionalpharmacologic variables in disguised form (Fisher,1970).

This notion is illustrated by many ongoingstudies of the endorphins, hormones, immuno-logical mechanisms and conditioned responsesand Wickramasekera, for example, has in-corporated current thinking on the placebo-response in his general definition of itas a 'composite of patterned interactingverbal-subjective, motor, neuro-endocrine, andneuro-immunological response systems that canattenuate or potentiate both the underlyingmechanisms of pathophysiology and overtclinical symptoms' (Wickramasekera, 1985).

Perhaps the most far-reaching extension ofthis standpoint has come from the perspective ofmolecular biology. Thus Ivor Black, who adoptsa weak reductionist position, goes so far as tomaintain that information processing within thenervous system involves the postulate thatsymbols are actually physical structures thatconstitute a neural language representing thephysical world (Black, 1991). This is renderedpossible because some molecules have multiplefunctions, acting simultaneously as physiologicalsignals and as symbols that carry environmentalinformation. On the basis of laboratory ex-periment Black is prepared to suggest that theneurobiology of the brain can be changed by thealteration of mental imagery that evokes at-tention, hypervigilance and anxiety. The high-level software, he asserts, is capable of modifyingthe low-level hardware of the brain. Here is anapproach to placebo action as a key element ofthe non-biological environment, which has stillto be explored by empirical research. It may bethat the 'humble humbug' may help define theboundaries between specific and non-specific,body and mind, soma and psyche.

REFERENCESAlexander, F. & Szasz, T. S. (1952). The psychosomatic approach in

medicine. In Dynamic Psychiatry (ed. F. Alexander and H. Ross),pp. 369-400. University of Chicago Press: Chicago.

Attardi, G. & Sperry, R. (1963). Preferential selection of centralpathways by regenerating optic fibers. Experimental Neurology 7,46-64.

Beecher, H. K. (1955). The powerful placebo. Journal of the AmericanMedical Association 159, 1602-1606.

Black, I. B. (1991). Information in the Brain: A Molecular Perspective.MIT Press: Cambridge, Mass.

Boyle, R. (1685). Of the Reconcileableness of Specifick Medicines tothe Corpuscular Philosophy. S. Smith: London.

Brill, H. & Patton, R. E. (1955). Analysis of population reduction inNew York State mental hospitals during the first four years oflarge-scale therapy with psychotropic drugs. American Journal ofPsychiatry 116, 495-508.

Brody, H. (1977). Placebos and the Philosophy of Medicine. Universityof Chicago Press: Chicago.

Brody, H. (1985). Placebo effect: an examination of Griinbaum'sDefinition. In Placebo: Theory, Research and Mechanisms (ed. L.White, B. Tursky and G. F. Schwartz), pp. 37-58. GuildfordPress: New York.

Brody, H. & Waters, D. B. (1980). Diagnosis is treatment. Journal ofFamily Practice 10, 445-449.

Bruner, J. (1990). Acts of Meaning, pp. 20-21. Harvard UniversityPress: Cambridge, Mass.

Cheney, C. O. & Drewry, P. H. (1938). Results of non-specifictreatment in dementia praecox. American Journal of Psychiatry95, 203-208.

Clark, A. J. (1937). General pharmacology. In Handbuch derexperimentellen Pharmakologie, Ergdmungswerk, vol. 4. (ed. A.Heffter). pp. 25-41. J. Springer: Berlin.

Cook, L. C. (1959). Comparison of abnormal behavioural statesinduced by psychotropic drugs in animals and man (Discussion).In Neuro-psychopharmacology (ed. P. Bradley, P. Deniker and C.Radouco-Thomas), p. 123. Elsevier: Amsterdam.

Downing, R. W. & Rickels, K. (1978). Non-specific factors and theirinteraction with psychological treatment in pharmacotherapy. InPsychopharmacology: A Generation of Progress (ed. M. A. Upton,A. Di Mascop and K. F. Killam), pp. 1419-1428. Raven Press:New York.

Eccles, J. (1959). Discussion in Neuro-Psychopharmacology (ed. P.Bradley, P. Deniker and C. Radrico-Thomas), pp. 57-61. Elsevier:Amsterdam.

Eisenberg, L. (1984). The ambiguities of psychosomatic medicine.Journal of Psychosomatic Obstetrics and Gynaecology 3, 115-127.

Elkin, I., Parloff, M. B., Hadley, S. W. & Autry, J. H. (1988a).NIMH treatment of depression collaborative research program.Archives of General Psychiatry 42, 305-316.

Elkin, I., Pilkonis, P. A., Docherty, J. P. & Sotsky, S. M. (1988A).Conceptual and methodological issues in comparative studies ofpsychotherapy and pharmacotherapy, II: nature and timing oftreatment effects. American Journal of Psychiatry 145, 1070-1076.

Elkin, I., Pilkonis, P. A., Docherty, J. P. & Sotsky, S. M. (1988c).Conceptual and methodological issues in comparative studies ofpsychotherapy and pharmacotherapy, I: active ingredients andmechanisms of change. American Journal of Psychiatry 145,909-917.

Fisher, R. A. (1930). The Genetic Theory of Natural Selection.Clarendon Press: Oxford.

Fisher, S. (1970). Non-specific factors as determinants of behaviouralresponse to drugs. In Clinical Handbook of Psychopharmacology(ed. R. I. Shader), pp. 17-39. Science House: New York.

Frank, J. D. (1983). The placebo is psychotherapy. Behavioural andBrain Sciences 6, 291-292.

Frank, J. D. (1986). Psychotherapy - the transformation of mean-ings. Journal of the Royal Society of Medicine 74, 241-246.

Frank, J. D. (1989). Non-specific aspects of treatment: the view of apsychotherapist. In Non-Specific Aspects of Treatment (ed. M.Shepherd and N. Sartorius), pp. 95-114. Huber: Toronto.

Galdston, I. (1941-2). The concept of the specific in medicine.Transactions of the College of Physicians of Philadelphia, 4th series,9, 25-34.

Garattini, S. & Mennini, T. (1986). Critical notes on the specificity ofdrugs in the study of metabolism and functions of brain mono-amines. Unpublished manuscript.

Griinbaum, A. (1981). The placebo concept. Behaviour Research andTherapy 19, 157-167.

Hahn, A. & Kleinman, A. (1983). Belief as pathogen, belief as

578 M. Shepherd

medicine: 'voodoo death' and the 'placebo phenomenon1 inanthropological perspective. Medical Anthropology Quarterly 14,3-19.

Harris, T. & Brown, G. W. (1989). The LEDS findings in the contextof other research: an overview. In Life Events and Illness (ed.G. W. Brown and T. O. Harris), pp. 57-94. Huber: Toronto.

Hill, A. B. (1971). Principles of Medical Statistics, 9th edn. p. 260.Lancet: London.

Hogben, L. (1954). The assessment of remedies. Medical Press, 13Oct. pp. 331-338.

Hogben, L. & Sim, M. (1953). The self-controlled and self-recordedclinical trial for low-grade morbidity. British Journal of PreventiveSocial and Medicine!, 163-179.

Joyce, C. R. B. (1989). Non-specific aspects of treatment from thepoint of a clinical pharmacologist. In Non-specific Aspects ofTreatment (ed. M. Shepherd & N. Sartorius), pp. 57-94. Huber:Toronto.

King, L. S. (1978). The Philosophy of Medicine, p. 98. UniversityPress: Cambridge, Mass.

Kraepelin, E. (1892). Uber die Beeinflussung einfacher psychischerVorgange durch einige Artzneimittel, p. 227. Fischer Verlag: Jena.

Lasagna, L. (1955). Placebos. Scientific American 193, 68-71.Lasagna, L. (1962). Some explored and unexplored psychological

variables in therapeutics. Proceedings of the Royal Society ofMedicine 55, 773-776.

Lewis, A. (1959). The impact of psychotropic drugs on the structure,function and future of psychiatric services, (a) In the hospitals. InNeuropsychopharmacology, vol. 1, (ed. P. B. Bradley, P. Deniker&C. Radouco-Thomas), pp. 207-211. Elsevier: Amsterdam.

Mayer-Gross, W. (1959). Model psychoses: their history, relevancyand limitations. American Journal of Psychiatry 115, 673-682.

Medawar, P. & Medawar, J. (1985). Aristotle to Zoos, p. 244. OxfordUniversity Press: Oxford.

Meehl, P. E. (1977). Specific etiology and other forms of stronginfluence: some quantitative meanings. Journal of Medicine andPhilosophy, 2, 33-53.

Muller, W. E. (1987). The Benzodiazapine Receptor: Drug Acceptoronly or a Physiologically Relevant Part of our Central NervousSystem? Cambridge University Press: Cambridge.

Needham, J. (1963). Discussion, in 'Scientific Change', ed. A. C.Crombie, p. 656. Heinemann: London.

0dcgaard, 0 . (1964). Pattern of discharge from Norwegian psy-chiatric hospitals before and after the introduction of psychotropicdrugs. American Journal Psychiatry 120, 772-778.

Pauling, L. (1940). A theory of the structure and process offormation of antibodies. Journal of the American Chemical Society62, 2643-2653.

Pepper, H. P. (1945). A note on the placebo. American Journal ofPharmacy 117,409-412.

Pickering, G. W. (1949). The place of the experimental method inmedicine. Proceedings of the Royal Society of Medicine XLI1,229-234.

Prioleau, L., Murdock, M. & Brody, No. (1983). An analysis ofpsychotherapy versus placebo studies. Behavioural and BrainSciences 6, 275-310.

Ryle, J. A. (1944). Social medicine: its meaning and its scope.Milbank Memorial Fund Quarterly 22, 58-63.

Sebeok, T. A. (1983). Psychotherapy and placebo. Behavioural andBrain Sciences 6, 300.

Shapiro, A. K. (1968). Semantics of the placebo. Psychiatric Quarterly42, 653-696.

Shepherd, M., Goodman, N. & Watt, D. C. (1961). The applicationof hospital statistics in the evaluation of pharmacotherapy in apsychiatric population. Comprehensive Psychiatry 2, 11-19.

Shryock, R. H. (1969). The concept of the specific in Americanmedicine. Proceedings of the 21st International Congress of theHistory of Medicine, pp. 755-761. Arti Grafiche & Cossidente:Rome.

Temkin, O. (1963). The scientific approach to disease: specific entityand individual sickness. In Scientific Change (ed. A. C. Crombie),pp. 629-647. Heinemann: London.

Temkin, O. (1972). Fernel, Joubert and Erastus on the Specificity ofCathartic Drugs. In Science, Medicine and Society in theRenaissance: Essays to Honor Walter Pagel(ed. A. G. Debus, vol.1, pp. 61-68. Neale Watson: New York.

Virchow, R. (1854). Specifiker und Specifisches. Virchow's Archive 6,3-33.

White, S. (1985). Medicine's humble humbug: four periods in theunderstanding of the placebo. Pharmacy in History 27, 1-10.

White, L., Thursky, B. & Schwartz, G. E. (eds.) (1985). Placebo:Theory, Research and Mechanisms. Guildford Press: New York.

Wickramasekera, I. (1985). A conditioned response model of theplacebo effect: predictions from the model. In Placebo: Theory,Research and Mechanisms (ed. L. White, B. Thursky and G. E.Schwartz), pp. 255-287. Guildford Press: New York.

Wolf, S. (1961). Disease as a way of life: neural integration insystemic pathology. Perspectives in Biology and Medicine IV,288-305.