In recent years labeling (or societal reac-tion) theory has aroused strong interestamong people concerned with mental ill-ness. From the perspective of labeling the-ory, the salient features of the behaviorpatterns called mental illness in countrieswhere Western psychiatry is practiced ap-pear to be as follows: (i) these behaviorsrepresent deviations from what is believedto be normal in particular sociocu!Yralgroups, (ii) the norms against which the de-viations are identified are different in dif-ferent groups, (iii) like other forms of de-viation they elicit societal reactions whichconvey disapproval and stigmatization, (iv)a label of mental illness applied to a personwhose behavior is deviant tends to becomefixed, (v) the person labeled as mentally illis thereby encouraged to learn and accept arole identity which perpetuates the stigma-tizing behavior pattern, (vi) individualswho are powerless in a social group aremore vulnerable to this process than othersare, and (vii) because social agencies inmodern industrial society contribute to thelabeling process they have the effect ofcreating problems for those they treatrather than easing problems.

This school of thought lemerged mainlywithin sociology, as an extension of studiesof social deviance in which crime and de-linquency were originally the major focus(1). It is also associated with psychiatrythrough, for example, Thqmas Szasz andR. D. Laing (2). These ideas have come tobe called a "sociologicgi model" of mentalillness, for they center jj learning and thesocial construction of norms. They beganto be formulated about 25 years ago (3),commanded growing attention in the late1960's, and have-been influential in recentmajor changes in public programs for psy-chiatric care, especially the deinstitutional-ization which is occurring in a number ofstates (4, 5).Several aspects of the theory receive

12 MARCH 1976

support from a study reported in Scienceby David Rosenhan (6), based on the expe-riences of eight sane subjects who gainedadmission to psychiatric hospitals, werediagnosed as schizophrenic, and remainedas patients an average of 19 days until dis-charged as "in remission." Rosenhan ar-gues that "we cannot distinguish insanityfrom sanity" (6, p. 257). He associates hiswork with "anthropological consid-erations" and cites Ruth Benedict (7) as anearly contributor to a theme he pursues,which is that "what is viewed as normal inone culture may be seen as quite aberrantin another" (6, p. 250). I: indicates thatthe perception of behavior as being schizo-phrenic is relative to context, for "psychi-atric diagnosis betrays little about thepatient but much about the environment inwhich an observer finds him." He arguesthat, despite the effort to humanize treat-ment of disturbed people by calling thempatients and labeling them mentally ill, theattitudes of professionals and the public atlarge are characterized by "fear, hostility,aloofness, suspicion, and dread." Once thelabel of schizophrenia has been applied,the "diagnosis acts on all of them"-patient, family, and relatives-"as a self-fulfilling prophecy. Eventually, the patienthimself accepts the diagnosis, with all of itssurplus meanings and expectations, andbehaves accordingly" (6, p. 254)-The research to be described here

presents an alternative perspective derivedfrom cross-cultural comparisons, mainlyof two widely separated and distinctly con-trasting non-Western groups, Eskimos ofnorthwest Alaska and Yorubas of rural,tropical Nigeria. It is concerned with themeanings attached to behaviors whichwould be labeled mental illness in our so-ciety. I interpret these data as raising im-portant questions about certain assump-tions in the labeling thesis and therefore ascasting doubt on its validity as a major ex-

planation of mental illness, especially withrespect to schizophrenia. These cross-cul-tural investigations suggest that relativismhas been exaggerated by labeling theoristsand that in widely different cultural andenvironmental situations sanity appears tobe distinguishable from insanity by cuesthat are very similar to those used in theWestern world.

The Labeling Orientation

As Edwin Schur (8) points out, if label-ing theory is conceived broadly it is the ap-plication of George Herbert Mead's theo-ries about self-other interactions to a defi-nition of social deviance extended to in-clude human problems ranging from crimeto blindness. Labeling theory emphasizesthe social meanings imputed to deviant be-havior and focuses on the unfolding pro-cesses of interaction whereby self-defini-tion is influenced by others. Further, "it isa central tenet of the labeling perspectivethat neither acts nor individuals are'deviant' in the sense of immutable, 'objec-tive' reality without reference to processesof social definition." Schur states that"'this relativism may be viewed as a majorstrength" of labeling theory (8, p. 14).\Edwin Lemert's concept of secondarydeviance (9) is of critical importance inlinking self-other considerations to devia-tions. Secondary deviation occurs when aperson learns the role and accepts the iden-tity of a deviant as the basis of his life-style. It is a response to a response; nega-tive feedback from significant others rein-forces and stabilizes the behavior that ini-tially produced it. Applied to criminality,this idea has created general awareness ofa process whereby a young person on beinglabeled a juvenile delinquent may enter anetwork of contingencies that lead ulti-mately to his learning criminal activitiesand "hardening" as a criminal rather thanto the correction of behavior.

In The Making of Blind Men, RobertScott points to a similar process regardinga very different type of deviance (10). If aperson is labeled blind by certain adminis-trative criteria he is likely to become en-meshed in care-giving agencies that en-courage him to accept a definition of him-self as helpless and to learn to play the roleof the blind man. These experiences mayeven inhibit the use of residual vision.Scott shows that institutions for the blindvary in the degree to which they encourageacceptance or rejection of the deviant roleand that these differences are related to

The author is associate professor of anthropology inthe Harvard School of Public Health, Boston, Massa-chusetts 02115. A portion of this article was presentedat the 1973 meeting of the Atlantic Provinces Psychiat-ric Association in Halifax, Nova Scotia.

1019

Psychiatric Labeling inCross-Cultural Perspective

Similar kinds of disturbed behavior appearto be labeled abnormal in diverse cultures.

Jane M. Murphy

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

3, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

differences in the life-style of blind men.Insofar as the labeling concept has beenemployed in this way I believe it is soundand has disclosed new and valuable infor-mation.The application of labeling ideas to

mental illness has tended to take a differ-ent course (11) and has aroused consid-erable controversy, as indicated, for ex-ample, in the continuing exchange betweenThomas Scheff and Walter Gove et al. (12-15). One question in this controversy iswhether mental illness should be consid-ered a "pure case" of secondary deviationor a more complex case. Lemert's formu-lation of the concept of secondary devia-tion was influenced by his investigation ofstuttering, and he suggests that stutteringrepresents the pure case: "Stuttering thusfar has defied efforts at causative ex-planation.... It appears to be exclusively aprocess-product in which, to pursue themetaphor, normal speech variations, or atmost, minor abnormalities of speech (pri-mary stuttering) can be fed into an inter-actional or evaluational process and comeout as secondary stuttering" (9, p. 56).The important point here is that primary

deviance is considered to be normal varia-tion or only "minor abnormalities," andthe influence of societal reactions is consid-ered genuinely causative. Societal reac-tions "work on" and "mold" normal vari-ations of speech to "create" stuttering. Formental illness the labeling theorists havetended to use the "pure case" model ratherthan the more complex model representedby blindness, where lack or loss of sight isprimary deviance and the role of blind manis secondary deviance.

Scheff has provided the most systematictheoretical statement regarding labelingand mental illness, and in his formulationthe primary deviations that are fed into in-teractional processing to come out as men-tal illness are described as "amorphous,""unstructured," and "residual" violationsof a society's norms (11, pp. 33, 82). Ro-senhan suggests that the behaviors labeledschizophrenic might be " 'sane' outside thepsychiatric hospital but seem insane in it... [because patients] are responding to abizarre setting" (6, p. 257). Lemert saysthat social exclusion can "create a para-noid disposition in the absence of any spe-cial character structure" (9, p. 198). Fur-ther, many have posited that behavior wecall mental illness might be considerednormal in a different culture or in a minor-ity social class. Thus, the primary devia-tions of mental illness are held to be for themost part insignificant, and societal reac-tions become the main etiological factor.

This view is reminiscent of ideas abouthuman plasticity, cultural determinism,and cultural relativism which were promi-

1020

nent in what used to be called the culture-personality studies of anthropology. Infact the influence of culture-personality onlabeling theory is explicitly stated by Le-mert, who was trained jointly in sociologyand anthropology and who has drawn onnon-Western studies throughout his career.The influence is equally acknowledged byRosenhan (6). It seems to me that numbersof proponents of labeling theory assumethat the expanding body of data from non-Western areas has supported the relativistpropositions put forth by Benedict andothers in the 1930's and '40's (16). Indeed,it was my own assumption when I begananthropological work with Eskimos. Ithought I would find their conception ofnormality and abnormality to be very dif-ferent, if not opposite, from that held inWestern culture. This did not prove to bethe case, and my experience is not unique.Anthropologists who have been conductingfield research in recent years using moresystematic methods but continuing towork on the relations between individualbehavior and cultural context tend to holda greatly modified view of the extent of in-dividual plasticity and the molding force ofculture (17, 18).

It would be misleading on my part toimply that all theory building and investi-gation regarding the relation of labelingand mental illness have followed the pure-case model. In their studies on mental re-tardation Robert Edgerton and Jane Mer-cer use moderate labeling ideas and showthat social reactions are related to differ-ences in the ways subnormal individualsare able to function both in and outside ofinstitutions (19). A growing number ofstudies of alcoholism, many of them influ-enced by labeling views, have demon-strated that social attitudes and the vari-able meanings attached to drinking arecorrelated with marked differences in alco-holism rates in various cultural groups(20). There are, in addition, numbers ofstudies of the social pathways leading tohospitalization, the impact of hospital-ization, attitudes toward discharged men-tal patients, and so on which reveal impor-tant outcomes for the mentally ill withoutimputing to societal reactions the degree ofsignificance given them in the more de-terministic formulations.Most labeling studies of mental illness

have been carried out in the United Statesand the United Kingdom. Variations in thedefinition and tolerance of mental illnesshave mainly been studied in groups at dif-ferent social class levels in industrializedsociety (21). Since cultural relativism isone of the main elements of the orienta-tion, it seems useful to put some of the bas-ic labeling questions to non-Western data.As background for this, I quote from four

contributors to labeling theory: Scheff, Er-ving Goffman, Theodore Sarbin, and Da-vid Mechanic. These references do not en-compass the breadth and elaboration ofeach contributor's own approach to theproblem of mental illness, but they do re-flect the view of cultural relativity whichruns throughout the labeling orientation.

Scheff says that "the culture of thegroup provides a vocabulary of terms forcategorizing many norm violations" (11,pp. 33, 82). These designate deviationssuch as crime and drunkenness. There is aresidual category of diverse kinds of devia--tions which constitute an affront to the un-conscious definition of decency and realityuniquely characteristic of each culture.Scheff posits that the "culture provides noexplicit label" for these deviations but theynevertheless take form in the minds of so-cietal agents as "stereotypes of insanity."When people around a deviant respond tohim in terms of these stereotypes, "hisamorphous and unstructured rule-break-ing tends to crystallize in conformity tothese expectations." Scheff further sug-gests that these cultural stereotypes tend to~produce uniformity of symptoms within acultural group and "enormous differencesiiith-e-iaiiifet symptoms mirtaldisorder between societies."

It has been pointed out that there ap-pears to be a contradiction in one aspect ofScheff's theory (12, p. 876; 22). It is diffi-cult to accept that a socially shared imageof behavior that can influence action andhas the concreteness of a stereotype shouldlack a name. It is possible Scheff meantthat in the evolution of language a label forinsanity was the last to emerge because itrefers to a residue of norm violations. Thedating of words is beyond the scope of thedata to be presented here, but it will bepossible to see whether an explicit labelcurrently exists in the two cultures studied,a hunting-gathering culture (Eskimo) andan agricultural society (Yoruba), neither ofwhich developed a written language. If aword for insanity occurs we can then inves-tigate the kinds of behaviors therein de-noted.

Regarding our own society, Goffmanstresses that the "perception of losingone's mind is based on culturally derivedand socially engrained stereotypes as to thesignificance of symptoms such as hearingvoices, losing temporal and spatial orienta-tion, and sensing that one is being fol-lowed" (23). He further indicates thatthere is cultural variation in this kind ofimagery and differential encouragementfor such a view of oneself. This makes itappropriate to ask whether hallucinations,delusions, and disorientations are presentor absent from the conception of losingone's mind in Yoruba and Eskimo cul-

SCIENCE, VOL. 191

tures, assuming they have a stereotype ofinsanity at all.

Labeling theorists express considerabledissatisfaction with the concept of mentalillness, pointing out that it is a vague andeuphemistic metaphor and ties togetherphenomena that are neither "mental" nor-"illness." They argue that mental illness isa myth developed in Western societies,that the term represents an abortive effortto improve the treatment of people pre-viously called lunatics, that in the name ofthis myth we continue to incarcerate, pun-ish, and degrade people for deviating fromnorms. Sarbin suggests that defining be-havioral aberrations as illness occurred inmedieval Europe as a way to relabel peoplewho might otherwise have been burned atthe stake as witches (24). He further sug-gests that it was during this phase of West-ern history that the concept of mind cameinto being. It was used as a way to explainperplexing behavior that could not be re-lated to occurrences external to the person.It is "as if there are states of mind" thatcause these patterns of conduct. The "asif" was transmuted into the myth that themind exists as a real entity and can there-fore be sick or healthy.

In the data to be given, it will be possibleto ask whether the idea of an inner statethat influences conduct is found in thesenon-Western groups and, since bothgroups believe in witchcraft, whether astereotype of insanity is associated with theconduct of witches. Everywhere that witch-craft has been systematically studied therole of the witch involves deviances thatare heavily censured. The witch carries outpractices that are believed to harm peoplethrough supernatural means. If the insaneperson and the witch are equated in the be-liefs of non-Western groups, it would ap-pear to follow that in those groups mentalillness is thought of as social deviance; andthis would be a telling point for labelingtheory.

Mechanic makes the point that "al-though seemingly obvious, it is importantto state that what may be viewed asdeviant in one social group may be toler-ated in another, and rewarded in still othergroups" (25). He emphasizes that the so-cial response may influence the frequencywith which the deviant behavior occurs. Ithas been hypothesized by a number of re-searchers that holy men, shamans, or witchdoctors are psychotics who have been re-warded for their psychotic behavior bybeing made incumbents of highly regardedand useful roles (26). This is the obverse ofthe possibility that the insane are thoughtof as witches. The role of the healer carriesgreat power and approval. The idea of so-cial rewards for mental illness underscoresthe lengths to which relativity can be car-

12 MARCH 1976

ried, for it suggests that the social defini-tion of one kind of behavior can turn it intosuch opposing roles as the defamed witchor the renowned shaman. Mechanic'spoints make it appropriate, therefore, toask whether the shamans in Eskimo cul-ture and the healers in Yoruba culture arethought by the people to be mentally illand whether the rates of such mental ill-ness in these groups are similar to or dif-ferent from those in the West.

Scheff, Goffman, Sarbin, and Mechanicshare the view that in our society the appel-lation "mentally ill" is a "stigmatizing"and "brutalizing" assessment. It robs theindividual of identity through profound"mortification" and suggests that he is a"non-person." It forces him into an as-cribed role, exit from which is extremelydifficult. Thus another question is posed: IfEskimos and Yorubas have a stereotype ofinsanity, are they less harsh than we withthose defined as insane?To illustrate the model I have in mind

for exploring these questions I will first de-scribe a non-Western event which suggeststhat certain aspects of labeling theory arevalid. It does not concern mental illnessbut it demonstrates the use of labels as ar-bitrary social definitions in the labelingtheory sense. The case is reported by W. H.R. Rivers in connection with his analysis ofthe concept of death among the Melane-sians (27):Some persons who are seriously ill and

likely to die or who are so old that from theMelanesian point of view they are ready todie are labeled by the word mate, whichmeans "dead person." They become there-by subjects of a ceremonial live burial. Itcan be argued that the Melanesians have aconcept of death which is a social fiction. Itembodies what they arbitrarily agree to de-fine as death and is a distortion of realityas seen by most cultural groups. The labelmate involves a degradation ceremony inwhich an elderly person is deprived of hisrights and is literally "mortified." He isperceived "as if dead" and then buried.The linguistic relativist might even say thatthis use of the word mate shows that theMelanesians do not perceive death bymeans of the indicators of vital functioningapplied in Western society (28).

Rivers's own conclusion is that the Mel-anesians view death the way we do and arecognizant of the difference between biolog-ical and social mate. Biological mate is byfar the commoner phenomenon. In theirpractice of live burials the Melanesians infact take close note of two typical pre-cursors of death-old age and illness.

It seems clear, however, that sociallysanctioned acts based on symbolic mean-ings, such as those involved in social mate,are powerful in influencing the course of

human affairs. They can be treacherouslyabused and lead to what we think of ascruel outcomes. Rivers says that the prac-tice is not conceived to be cruel or degrad-ing by the Melanesians because in theirmeaning the burial relieves the pe-rson of aworn-out earth-life so that he can enter thehigher status of the spiritual afterlife. Byour standards the Melanesian inter-pretation would nevertheless be considereda collective rationalization of "geronti-cide." Whatever the intent, the socially de-fined death of elderly Melanesians is amyth and serves as a model of what I un-derstand the labeling theorists to mean bythe "myth of mental illness." Thus a finalquestion: Do the Eskimos and Yorubassubscribe to such fictions about mental ill-ness through which they perpetrate in-humanity and degradation?

Method of Study

The data to be presented derive mainlyfrom a year of field work, in 1954-55, in avillage of Yupik-speaking Eskimos on anisland in the Bering Sea, and an investiga-tion of similar length, in 1961 and 1963,among Egba Yorubas. I also draw onshorter periods of field work in Gambia,Sudan, and South Vietnam.Some of the Eskimo data came from a

key informant, who systematically de-scribed the life experiences of the 499 Eski-mos who constituted a total village censusover the 15 years previous to and includingthe year of investigation. In addition, a dic-tionary of Eskimo words for illness and de-viance was developed. Extended life histo-ries of a small number of Eskimos weregathered. Also daily observations andcomments from Eskimos about Eskimos(both in their own village and in otherareas known to them) were recorded forthe purpose of understanding their concep-tions of behavior (29).The approach among the Yorubas was

different in that I worked with a group ofthree native healers and a member of an in-digenous cult. Interviews were directedtoward understanding Yoruba concepts ofbehavior in the abstract and centered onactual people only to the extent that ac-quaintances and patients were brought intothe discussion as illustration (30).The Eskimo data served as the base for

an epidemiological study of the village in1955, and the Yoruba data constituted oneof the first phases of a larger epidemiolog-ical study carried out with a group of Ni-gerian and U.S. colleagues in which westudied 416 adults, of whom 245 consti-tuted a representative sample from 14 vil-lages (31).

In The Social Meanings of Suicide, a1021

study affiliated with the labeling tradition,Jack Douglas has shown the weakness ofofficial statistics as a basis for judging thesocial significance of behavioral phenome-na in groups (32). The Eskimo and Yorubastudies reflect a similar orientation aboutthe inadequacies of mental hospital statis-tics for the purposes at hand. As has beendone in many labeling studies, I relied onparticipant observation and interviewingabout microcultural events. The focus wason indigenous meanings. These meaningswere then used as a basis for counting sim-ilar behavior patterns, so that they weredefined from within a cultural group ratherthan by imposed criteria.

In these studies I have considered lan-guage to be the main repository of labels.Insofar as there is a counterpart to the offi-cial recognition of mental illness involvedin hospital commitment in a Western so-ciety, it resides in what Eskimos and Yor-ubas say are the kinds of people treated byshamans and native healers.

Labeled Behavior Patterns

The first specific question is: Do Eski-mos and Yorubas have labels for psycho-logical and behavioral differences that bearany resemblance to what we mean by men-tal illness? These groups clearly recognizedifferences among themselves and describethese in terms of what people do and whatthey say they feel and believe. Some of thedifferences lead people to seek the aid ofhealers and some do not, some differencesarouse sympathy and protection while oth-ers arouse disapproval, some are calledsickness and others health, some are con-sidered misconduct and others good con-duct. Some are described by a single wordor nominative phrase. Some that seem tohave common features are described invarying circumlocutions and sentences. Ifa word exists for a complex pattern of be-havior it seems acceptable to assume thatthe concept of that pattern has been crys-tallized out of a welter of specific attributesand that the word qualifies as an explicitlabel.Of major importance is whether or not

the Yorubas and Eskimos conceptualize adistinction between body and mind and at-tribute differences in functioning to one orthe other. The first indication of such a dis-tinction arose early in the Eskimo censusreview when a woman was described inthese terms: "Her sickness is getting wildand out of mind ... but she might have hadsickness in her body too." The Eskimoword for her was nuthkavihak. It becameclear from other descriptions that the wordrefers to a complex pattern of behavioralprocesses of which the hallmark is con-

1022

ceived to be that something inside the per-son-the soul, the spirit, the mind-is outof order. Descriptions of how nuthkavihakis manifested include such phenomena astalking to oneself, screaming at someonewho does not exist, believing that a child orhusband was murdered by witchcraft whennobody else believes it, believing oneself tobe an animal, refusing to eat for fear eatingwill kill one, refusing to talk, runningaway, getting lost, hiding in strange places,making strange grimaces, drinking urine,becoming strong and violent, killing dogs,and threatening people. Eskimos translatenuthkavihak as "being crazy."

There is a Yoruba word, were, which isalso translated as insanity. The phenomenainclude hearing voices and trying to getother people to see their source thoughnone can be seen, laughing when there isnothing to laugh at, talking all the time ornot talking at all, asking oneself questionsand answering them, picking up sticks andleaves for no purpose except to put them ina pile, throwing away food because it isthought to contain juju, tearing off one'sclothes, setting fires, defecating in publicand then mushing around in the feces, tak-ing up a weapon and suddenly hittingsomeone with it, breaking things in a stateof being stronger than normal, believingthat an odor is continuously being emittedfrom one's body.

For both nuthkavihak and were in-digenous healing practices are used. Infact, among the Yorubas some native heal-ers specialize in the treatment of were (33,34).The profile of were behaviors is based

not only on what the healers described inthe abstract but also on data concerningtwo members of the sample identified aswere by the village headman and a groupof 28 were patients in the custody of nativehealers and in a Nigerian mental hospital.The profile of nuthkavihak is built from in-formation about four individuals withinthe 15-year population of 499 persons andsix Eskimos from earlier times and from arelated Eskimo settlement in Siberia.Of paramount significance is the fact

that were and nuthkavihak were never usedfor a single phenomenon such as hearingvoices, but rather were applied to a patternin which three or four of the phenomenadescribed above existed together. It istherefore possible to examine the situ-ations in which a person exhibited one oranother of the listed behaviors but was notlabeled insane.The ability to see things other people do

not see and to look into the future andprophesy is a clearly recognized and highlyvalued trait. It is called "thinness" by Es-kimos. This ability is used by numerousminor Eskimo diviners and is the out-

standing characteristic of the shaman. Thepeople called "thin" outnumber thosecalled insane by at least eight to one.Moreover, there were no instances when a"thin" person was called nuthkavihak.When a shaman undertakes a curing rite

he becomes possessed by the spirit of ananimal; he "deludes" himself, so to speak,into believing that he is an animal. Consid-er this description (35):

The seance is opened by singing and drumming.After a time the shamaness falls down very hardon the floor. In a while, the tapping of her fin-gers and toes is heard on the walrus skin floor.Slowly she gets up, and already she is thought to"look awful, like a dog, very scary." She crawlsback and forth across the floor making growlingsounds. In this state she begins to carry out thevarious rites which Eskimos believe will curesickness, such as sucking the illness out of thebody and blowing it into the air. Following thisthe shamaness falls to the floor again and theseance is over.

Compare this to the case, reported byMorton Teicher, of a Baffin Island Eskimowho believed that a fox had entered herbody (36). This was not associated withshamanizing but was a continuous belief.She barked herself hoarse, tried to clawher husband, thought her feet were turninginto fox paws, believed that the fox wasmoving up in her body so that she couldfeel its hair in her mouth, lost control ofher bowels at times, and finally became soexcited that she was tied up and put into acoffin-like box with an opening at the headthrough which she could be fed. This wom-an was thought to be crazy but the sha-maness not. One Eskimo summarized thedistinction this way: "When the shaman ishealing he is out of his mind, but he is notcrazy." Figure 1 is a picture selected by anEskimo to illustrate the shaman's appear-ance during a seance (37).

This suggests that seeing, hearing, andbelieving things that are not seen, heard,and believed by all members of the groupare sometimes linked to insanity andsometimes not. The distinction appears tobe the degree to which they are controlledand utilized for a specific social function.The inability to control these processes iswhat is meant by a mind out of order;when a mind is out of order it will not onlyfail to control sensory perception but willalso fail to control behavior. Another Eski-mo who was asked to define nuthkavihaksaid that it means "the mind does not con-trol the person, he is crazy." I take this tomean that volition is implicated, that hear-ing voices, for example, can be voluntaryor involuntary, and that it is mainly the in-voluntary forms that are associated withwere and nuthkavihak.

In cultures such as Eskimo and Yoruba,where clairvoyant kinds of mental phe-nomena are encouraged and preternatural

SCIENCE, VOL. 191

experiences are valued, something similarto what we might call hallucinations anddelusions can probably be learned or simu-lated. A favorable audience reaction islikely to stabilize the performance of thepeople who fill the roles of fortune-tellerand faith healer. For example, the shaman-ess described above was unable to keep herpatient alive but her performance was con-sidered to have been well executed; she wassaid to have done "all her part, acting likea dog." The Eskimos believe that a personcan learn to be a shaman. Their view ofnuthkavihak is something that befalls theperson, a pattern of behavioral processesthat can appear and disappear, lasting along time with some people and a shorttime with others.A number of researchers in the field of

cross-cultural psychiatry take the positionthat the underlying processes of insanityare the same everywhere but that their spe-cific content varies between cultural groups(38). A psychotic person, it is thought,could not make use of the imagery ofChrist if he had not been exposed to theChristian tradition and he could not elabo-rate ideas about the wittiko cannibalisticmonster if not exposed to Cree and OjibwaIndian traditions (39). It would seem thatif a culture-specific stereotype of the con-

tent of psychosis exists in a group it mighthave the kind of influence suggested in la-beling theory. If the content stereotypewere applied to the unstructured delusionsof a psychotic his thought productionsmight be shaped and stabilized around thetheme of that stereotype.There have been several attempts to

study phenomena such as wittiko and pib-loktoq, the former being thought of as theculturally defined content of a psychoticprocess in which the person believes him-self to be a cannibalistic monster and thelatter as a culture-specific form of hysteriafound in the arctic (40, p. 218; 41). The evi-dence of their existence comes from earlyethnographies. It has been difficult in thecontemporary period to locate people whohave these illnesses (42). If the availabilityof a content stereotype has the effect onewould expect from labeling theory, thestereotype should have sustained the pat-tern, but in fact these content patternsseem to have disappeared.

Prominent in the descriptions of the im-ages and behavior of people labeled wereand nuthkavihak were cultural beliefs andpractices as well as features of the naturalenvironment. Eskimo ideation concernedarctic animals and Eskimo people, objects,and spirits. The Yoruba ideation was based

on tropical animals and Yoruba figures.The cultural variation was, in other words,general. There was no evidence that if aperson were to become were or nuthkavi-hak he would reveal one specific delusionbased on cultural mythology. In this re-gard I reach the same conclusion as RogerBrown did when he set out to see how farlabeling ideas would aid his understandingof hospitalized schizophrenics: "Delusionsare as idiosyncratic as individual schizo-phrenics or normals.... There seems to benothing like a standard set of heresies, butonly endless variety" (15, p. 397).The answer to the first specific question,

whether Eskimos and Yorubas have labelsfor psychological and behavioral differ-ences resembling what we call mental ill-ness, is to my mind a definite yes. The ex-panding ethnographic literature on thistopic indicates that most other non-West-ern groups also have such labels [in addi-tion to the papers already cited see (43)].From this broad perspective it appears that(i) phenomenal processes of disturbedthought and behavior similar to schizo-phrenia are found in most cultures; (ii)they are sufficiently distinctive and notice-able that almost everywhere a name hasbeen created for them; (iii) over and abovesimilarity in processes, there is variability

Fig. 1 (left). The shaman during seance: "he isout of mind but not crazy." Fig. 2 (right).A man living on an abandoned anthill: "he isout of mind and crazy."

12 MARCH 1976 1023

in content which in a general way is col-ored by culture; and (iv) the role of socialfictions in perceiving and defining the phe-nomena seems to have been very slight.

Unlabeled Behavior Patterns

The questions of this section are: Dophenomena labeled mental illness by us go

unlabeled elsewhere, And if so what are theconsequences? Are there natural experi-ments of culture which allow us to gainsome understanding of the effects of not la-beling? From the linguistic relativist'sviewpoint, if phenomena are not namedthey are screened out of the perception ofthe people who speak that language; thusnot only would mental illness go unrecog-

nized if unlabeled but also the negative ef-fects of labeling could not pertain.

Although one cannot speak of mentalillness without reference to insanity andpsychoses, most people in our culturemean more by the term and include some

or all of the phenomena described in a text-book of psychiatry. Elsewhere I havepresented data about Eskimo and Yorubaterms, lack of terms, and levels of gener-

alization for mental retardation, con-

vulsions, and senility (30, 44). According tothe healers with whom I worked, the Yor-ubas have no word for senility but theyrecognize that some old people become in-capable of taking care of themselves, talkto themselves, are agitated, wander away

and get lost. In such cases they are

watched, fed, and protected in much thesame way as might be done in a nursinghome. The lack of an explicit label seems

to make little difference in how they are

treated.In contemporary Western society psy-

choneurotic patterns are thought of as one

of the main types of mental illness, yet neu-

rosis has a minor role in the labeling theo-ry literature (45). Since labeling theory isaddressed to the concept of mental illnessper se, one feels it ought to apply to theneurotic as well as the psychotic.

In working with the Eskimos and Yoru-bas I was unable to find a word that couldbe translated as a general reference to neu-

rosis or words that directly parallel our

meaning of anxiety and depressiorl. On theother hand, their words for emotional re-

sponses that we might classify as manifes-tations of anxiety or depression constitutea very large vocabulary. The Yoruba lexi-con includes, for example, words for unrestof mind which prevents sleep, being terri-fied at night, extreme bashfulness which islike a sense of shame, fear of being amongpeople, tenseness, and overeagerness. TheEskimo terms are translated as worryingtoo much until it makes the person sick,

1024

too easy to get afraid, crying with sadness,head down and rocking back and forth,shaking and trembling all over, afraid tostay indoors, and so on. The point is thatneither group had a single word or explicitlabel that lumped these phenomena togeth-er as constituting a general class of illnessby virtue of their underlying similarities or

as a pattern in which several componentsare usually found in association (46). In theterms of this article, these symptoms are

unlabeled but they do exist. People recog-

nize them and try to do something aboutthem. Some of them are conceived as se-

verely disabling and cause people to giveup aspects of their work (such as beingcaptain of a hunting boat); others appear

to be less serious. Some of them are tran-sient; others are life-long characteristics.Of special significance to the problem at

hand is the fact that most of these emo-

tional phenomena are definitely thought ofas illnesses for which the shaman and witchdoctor have effective cures. The number ofpeople who exhibit these phenomena isconsiderably in excess of those labeledwere and nuthkavihak. Among the Yoru-bas the ratio is approximately 12 to I andamong the Eskimos 14 to 1. In the clienteleof a typical shaman or healer a large pro-

portion would be people who came withsymptoms such as "unrest of mind thatprevents sleep" or "shaking and tremblingall the time."The answer to the question whether phe-

nomena we label mental illness go unla-beled elsewhere is thus also yes. These Es-kimos and Yorubas point out a large num-ber of psychological and behavioral phe-nomena which we would call neuroses butwhich they do not put together under sucha rubric. The consequence is not, however,a reduction in the number of persons whodisplay the phenomena or great differencein how they are treated: The fact that thesepeoples cannot categorically define some-

one as "a neurotic" or that the Yorubas donot talk about "a senile" appears mainlyto be a classification difference, and I amled to conclude that the phenomena existindependently of labels.

Evaluation of Behavior Patterns

Do non-Western groups evaluate the la-beled behaviors of mental illness negative-ly or positively? Are they more tolerant ofdeviance than we are? I shall consider firstthe related institutional values of the cul-ture, its roles and ceremonies, and then thenoninstitutionalized actions and attitudestoward the mentally ill.As pointed out earlier, it has been pro-

posed that the shaman role is a social nichein which psychopathology is socially useful

and that therefore mental disorder is posi-tively valued. Since the Eskimos do not be-lieve the shaman is nuthkavihak, it cannotbe insanity that invests the role with pres-tige in their eyes. It could be, however, thatsome other form of mental illness, possiblya neurotic disorder like hysteria, is consid-ered essential to what a shaman does andtherefore is accorded the same respect thatthe role as a whole commands.Among the 499 Eskimos 18 had shama-

nized at some time in their lives. None wasthought to be nuthkavihak. No other per-sonality characteristic or emotional re-sponse was given as typical of all of them,and in these regards the shamans seemedto be a random sample of the whole. Theonly feature I was able to determine ascommon to the group was that theyshamanized, and they did that with vari-able success.The Yoruba healer has not been de-

scribed in the literature as a mentally illperson, though some of the Yoruba healingcults consist of individuals who have beencured and thereafter participate in curingothers. The healers known to me and myconversations with Yorubas about theirhealers gave no evidence that mental ill-ness was a requisite. Thus as far as thegroups reported here are concerned, men-tal illness does not appear to be veneratedin these roles. If the shaman is to be con-sidered either psychotic or hysterical itseems to require that a Western definitionbe given to the portion of behavior specificto shamanizing.

If not institutionalized in an esteemedrole, is mental illness institutionalized in acontemptible role? Both the Yorubas andthe Eskimos have a clearly defined role ofwitch as the human purveyor of magicallyevil influences. Though feared, the man orwoman who is believed to use magic in thisway is held in low esteem.

Is insanity or other mental illness primafacie evidence that a person is a witch? Ifone tries to answer this by identifying thepeople labeled were or nuthkavihak andthen the people labeled witches and com-paring the two groups to see how muchthey overlap in membership, as I did re-garding the shamans, a serious problemarises. The difficulty is in identifying thewitches. Unlike shamanizing, which is apublic act, the use of evil magic is ex-ceedingly secretive. I did note, however,that there was no correspondence betweenthe group of Eskimos said to have been in-sane at some point in their lives and the sixpeople named as auvinak (witch) by atleast one Eskimo.

In the more generalized informationfrom the Yoruba healers it was evidentthat insanity was often believed to resultfrom the use of evil magic but an insane

SCIENCE, VOL. 191

person was rarely believed to use it againstothers. Thus my interpretation of whethermental illness is built into the role of witchis similar to the view presented about therole of healer. Some insane people haveprobably been accused of being witches,but it has been by happenstance, not be-cause witching and insanity are consideredto be the same thing and equally stigma-tized.

In these regards the Eskimos and Yoru-bas seem to have much in common withthe Zapotecs studied by Henry Selby (47,pp. 41-42). His work focuses on witchcraftas a major form of social deviance, and heinterprets his information as supporting la-beling ideas (48), especially the vulnerabil-ity of outsiders to labeling. He found thataccusations of witchcraft are more likelyto be leveled at someone outside the imme-diate group of kin and neighbors than at agroup member. However, after "talkingabout deviance for months" with his Zapo-tec informants, Selby realized that he hadno information on mental illness. He ex-plored this topic separately and "found outthat there were people who were 'crazy'"and that the condition was defined as hav-ing "something to do with the soul and wassymptomized by agitated motor behavior,ataraxia, violent purposeless movement,and the inability to talk in ways that peoplecould readily understand." Clearly, theZapotecs have a conception of insanity,and, like Eskimos and Yorubas, they donot classify it in the same frame of refer-ence with such norm transgressions aswitching, envy, stinginess, and adultery.Another way in which a culture might

institutionalize a negative view of mentalillness is through a degradation ceremonyor ritual slaying, as in the case of the Mela-nesian social mate. Ceremony is a pre-servative of custom, and there is volumi-nous information on ceremonies for heal-ing, ceremonies for effecting fertility ofland, animals, and humans, and rites ofpassage, as well as ceremonies in whichvarious forms of human sacrifice are car-ried out.

In view of the wide elaboration of cus-toms whereby groups of people enact theirnegative and positive values, it is perhapssurprising that no groups seem to have de-veloped the idea of ceremonially killing aninsane person in the prime of life just be-cause he is insane. Infanticide has some-times been conducted when a child wasborn grossly abnormal in a way whichmight later have emerged as brain damage,and senility may have been a contributingfactor in live burials. Also there is no doubtthat insane people have sometimes beendone away with, but that is different fromritual sacrifice. There is no evidence as faras I can determine that killing the insane12 MARCH 1976

has ever been standardized as a custom.There are, on the other hand, numerous in-dications from non-Western data that theceremony appropriate for people labeledmentally deranged is the ceremony of heal-ing (34, 36, 38, 43). Even the word "luna-tic" associates the phenomena with heal-ing, since it was usually the healer who wasbelieved to have power over such cosmicforces as the lunar changes which werethought to cause insanity.

Regarding informal behavior and atti-tudes toward the mentally ill it is difficultto draw conclusions, because there is evi-dence of a wide range of behaviors that canbe conceptualized as audience reactions.Insane people have been the objects of cer-tain restrictive measures among both theEskimos and the Yorubas. The Eskimosphysically restrain insane people in violentphases, follow them around, and forcethem to return home if they run away; andthere is one report of an insane man'sbeing killed in self-defense when, after kill-ing several dogs, he turned on his family.In describing the Chukchee, a Siberiangroup known to these Bering Strait Eski-mos, Waldemar Bogoras reports the caseof an insane woman who was tied to a poleduring periods of wildness (49). Teicher de-scribes, in addition to the coffin-like boxmentioned earlier, the use of an igloo withbars across the opening through whichfood could be passed (36). This is againsimilar to Selby's observations of Zapotecswho barred the door of a bamboo hut as away of restraining a psychotic man (47).The Yoruba healer of were often has 12

to 15 patients in custody at one time. Notinfrequently he shackles those who are in-clined to run off, and he may use variousherbal concoctions for sedation. In Niger-ia, where population is much denser thanin the arctic, it was not uncommon to seewere people wandering about the citystreets, sometimes naked, more oftendressed in odd assortments of tatteredclothing, almost always with long, dirt-la-den hair, talking to themselves, picking upobjects to save. In studying a group of suchvagrant psychotics Tolani Asuni notedthat they usually stayed in one locale, thatpeople fed them generously, allowed themto sleep in the market stalls, teased themmildly or laughed at them for minor devia-tions, and took action to control them onlyif the psychotics became violent (50).A case I encountered in Gambia illus-

trates the complexities of the situation andindicates that compassion and rejectionare sometimes both engaged. The case is ofa man, identified as insane, who lived some500 yards outside a village. The villagerslived in thatched mud houses. The mad-man lived on an abandoned anthill. It wasabout 2.5 meters long and 1.5 meters high

and the top had been worn away to matchthe contours of his body (Fig. 2). Exceptfor occasional visits to the village, he re-mained on this platform through day andnight and changing weather. His behaviorwas said to have become odd when he wasa young man, and when I saw him he hadnot spoken for years, although he some-times made grunting sounds. In one sensehe was as secluded and alienated from hissociety as patients in back wards are inours. On the other hand, the villagers al-ways put food out for him and gave himcigarettes. The latter act was accompaniedby laughter, because the insane man had acharacteristic way of bouncing severalleaps into the air to get away from anyonewho came close to him, and that was con-sidered amusing. Once a year someonewould forceably bathe him and put newclothes on him.

If one defines intolerance of mental ill-ness as the use of confinement, restraint, orexclusion from the community (or allow-ing people to confine or exclude them-selves), there does not appear to be a greatdeal of difference between Western andnon-Western groups in intolerance of thementally ill. Furthermore, there seems tobe little that is distinctively cultural in theattitudes and actions directed toward thementally ill, except in such matters as thatan abandoned anthill could not be used asan asylum in the arctic or a barred igloo inthe tropics. There is apparently a commonrange of possible responses to the mentallyill person, and the portion of the rangebrought to bear regarding a particular per-son is determined more by the nature of hisbehavior than by a preexisting cultural setto respond in a uniform way to whatever islabeled mental illness. If the behavior in-dicates helplessness, help tends to be given,especially in food and clothes. If the behav-ior appears foolish or incongruous (in thelight of the distinctive Eskimo and Yorubaviews of what is lg4norous), laughter is theresponse. If the Peh4vior is noisy and agi-tated, the respons' may be to try to quiet,sometimes by hers nd sometimes by oth-er means. If the behavior is violent orthreatening, the re,ponse is to restrain orsubdue.The answer to the question posed at the

beginning of this section seems to be thatthe patterns these groups label mental ill-ness (were or nuthkavihak) are not eval-uated in either a starkly positive or starklynegative way. The flavor and variability ofthe audience reactions to mental illnesssuggest the word "ambivalence." Two re-cent studies in the United States also in-dicate that stigma is not automatically anduniversally applied to mental illness andthat complex responses are typical in oursociety as well (51).

1025

Norm Violations

If these Eskimos and Yorubas are am-bivalent about mental illness, do theystrongly condemn any behaviors at all?Both groups have words for theft, cheating,lying, stinginess, drunkenness, and a largenumber of other behaviors which they con-sider to be specific acts of bad conduct.These, like the practice of witchcraft, arethought of as transgressions against socialstandards and are negatively sanctioned.

In addition, the Eskimos have a word,kunlangeta, which means "his mind knowswhat to do but he does not do it." This isan abstract term for the breaking of manyrules when awareness of the rules is not inquestion. It might be applied to a manwho, for example, repeatedly lies andcheats and steals things and does not gohunting and, when the other men are out ofthe village, takes sexual advantage ofmany women-someone who does not payattention to reprimands and who is alwaysbeing brought to the elders for punishment.One Eskimo among the 499 was calledkunlangeta. When asked what would havehappened to such a person traditionally, anEskimo said that probably "somebodywould have pushed him off the ice whennobody else was looking." This suggeststhat permissiveness has a limit even in acultural group which in some respects,such as attitude toward heterosexual activ-ity, is very lenient. The Yorubas have asimilarly abstract word, arankan, whichmeans a person who always goes his ownway regardless of others, who is uncooper-ative, full of malice, and bullheaded.

There are parallels between kunlangetaand arankan and our concept "psycho-path"-someone who consistently violatesthe norms of society in multiple ways.Also, some of the specific acts ofwrongdoing which Eskimos and Yorubasrecognize might in our society be calledevidence of "personality disorders." InWestern psychiatry, this term refers to sex-ual deviations, excessive use of drugs or al-cohol, and a variety of behaviors that pri-marily cause trouble for other people rath-er than for the doer.

It is of considerable interest that kunlan-geta and arankan are not behaviors thatthe shamans and healers are believed to beable to cure or change. As a matter of fact,when I pressed this point with the Yorubahealers they specifically denied that thesepatterns are illness. Both groups, however,believe that specific acts of wrongdoingmay make an individual vulnerable to ill-ness or other misfortune. For example, Es-kimos hold to a hunting ethic which pre-scribes ownership and sharing of animals;cheating in reference to the hunting code isthought of as a potential cause of physicalor mental illness. The social codes amongthe Yorubas are somewhat different, butthey also believe that breaking taboos cancause illness. It has been recognized by an-thropologists for nearly half a century thatamong peoples who believe in magic thereis remarkable similarity in the ex-planations of illness, and that trans-gression as well as witchcraft ranks high inthe accepted etiology of many non-West-ern groups (52). Believing that trans-gression causes illness is nevertheless quite

Table 1. Compilation of prevalence rates for schizophrenia, from Dunham (53).

Cases

Investigator Date Place Population RateNo. per

1000

Brugger 1929 Thuringia, Germany 37,546 71 1.9Brugger 1930-31 Bavaria, Germany 8,628 22 2.5StrQmgren 1935 Bornholm, Denmark 45,930 150 3.3Kaila 1936 Finland 418,472 1,798 4.2Bremer 1939-44 Northern Norway 1,325 6 4.5Sjogren 1944 Western Sweden 8,736 40 4.6Book 1946-49 Northern Sweden 8,931 85 9.6Fremming 1947 Denmark 5,500 50 9.0Essen-Moller 1947 Rural Sweden 2,550 17 6.6Mayer-Gross 1948 Rural Scotland 56,000 235 4.2Uchimura 1940 Hachizo, Japan 8,330 32 3.8Tsugawa 1941 Tokyo, Japan 2,712 6 2.2Akimoto 1941 Komoro, Japan 5,207 11 2.1Lin 1946-48 Formosa, China 19,931 43* 2.1Cohen and

Fairbank 1933 Baltimore, U.S. 56,044 127 2.3Lemkau 1936 Baltimore, U.S, 57,002 158 2.9Roth and Luton 1938-40 Rural Tennessee, U.S. 24,804 47 1.9Hollingsheadand Redlich 1950 New Haven, U.S. 236,940 845t 3.6

Eaton and Weil 1951 Hutterites, U.S. 8,542 9* 1.0*Inactive as well as active cases. tCases treated six months or more.

1026

different from believing that transgressionis illness.Thus the answer to the question of this

section appears to be that these groups dohave strong negative sanction for a numberof behaviors. A difference between theiropinions and those embodied in Westernpsychiatry is that the Eskimos and Yoru-bas do not consider these transgressionssymptomatic of illness or responsive to thetechniques used for healing.

Prevalence

Is the net effect of a non-Western way oflife such that fewer people suffer fromsomething they label mental illness than isthe case in the West? In view of the focuson were and nuthkavihak, attention willmainly be directed to this pattern of behav-ior and it will be compared with schizo-phrenia.There are available now a number of

epidemiological studies of mental illness indifferent countries and cultures. WarrenDunham has compared prevalence ratesfor schizophrenia from 19 surveys in Eu-rope, Asia, and North America; Table 1 isadapted from tables he presents (53). Likeseveral others who have studied these fig-ures, Dunham concludes that the preva-lence rates "are quite comparable" despitethe fact that some are based on hospitaldata and some on population surveys, de-spite differences in definitions and meth-ods, and despite the cultural variation in-volved.The rates of were and nuthkavihak can

be compared to rates of schizophrenia intwo Western surveys, one in Sweden andone in Canada. The Swedish study wascarried out by Erik Essen-Moller and col-leagues in two rural parishes for which apopulation register existed. Each memberof the population was interviewed by a psy-chiatrist. A prevalence rate of schizophre-nia is reported, with figures for cases in thecommunity and cases in a hospital during aspecific year (54). This design is similar tothe one I used among the Eskimos, where acensus register provided the base for deter-mining the population, and each personwas systematically described by at leastone other Eskimo. Focusing on the peopleliving in the specified year reduces the Es-kimo population studied from 499 to 348.The Canadian study, in which I was one

of the investigators, was based on a proba-bility sample of adults in a rural county(55). We designed the Yoruba study to ex-plore the possibilities of comparing mentalillness rates, and so used similar samplingprocedures. The rates in these two surveysare based on compilations of interview

SCIENCE, VOL. 191

data with selected respondents as well assystematic interviews about those respond-ents with local physicians in Canada andlocal village headmen in Nigeria.The results of comparing these studies is

that the proportion of people who exhib-ited or had at some time exhibited the pat-tern of behavior called schizophrenia, were,or nuthkavihak appears to be much thesame from group to group (Table 2). At thetime these studies were carried out, mentalhospitals existed all over the world. TheCanadian and Swedish populations aresimilar to the United States in having a siz-able number of large mental hospitals. TheEskimo population was considered to be inthe catchment area served by a mental hos-pital in the United States, and the Yorubavillages were in the vicinity of two mentalhospitals (56). For the Canadian and Yoru-ba studies we do not know the number ofpeople who might otherwise have been inthe communities but were hospitalizedduring the period when prevalence was sur-veyed. The Swedish and Eskimo studies,by virtue of starting with census registers,provide information on this point. The age-adjusted prevalence rate in the Swedishsurvey is 8.1 per 1000 when hospitalizedschizophrenics are included and the Eski-mo rate of nuthkavihak is increased to 8.8when the one hospitalized case is added.The number of schizophrenics, were,

and nuthkavihak in a population is small,but this comparison suggests that the ratesare similar. With a broader definition ofmental illness which I have explained else-where (it includes the neurotic-appearingsymptoms, the senile patterns, and so on)the total prevalence rates for the threegroups I have studied are: Canadian, 18percent; Eskimo, 19 percent; and Yoruba,15 percent (57).The answer to the last question above

seems thus to be that the non-Western wayof life does not offer protection againstmental illness to the point of making amarked difference in frequency. The ratesof mental illness patterns I have discussedare much more striking for similarity fromculture to culture than for difference. Thissuggests that the causes of mental illness,whether genetic or experiential, are ubiqui-tous in human groups.

Summary and Conclusions

Labeling theory proposes that the con-cept of mental illness is a cultural stereo-type referring to a residue of deviancewhich each society arbitrarily defines in adistinct way. It has been assumed that in-formation from cultures that are markedlydifferent from Western society supports12 MARCH 1976

Table 2. Rates of nonhospitalized schizophreniain two Western samples and of indigenously de-fined insanity in two non-Western samples.Rates are per 1000 population after adjustmentby the Weinberg method (58).

Cases

Group Date Size RateNo. per

1000

Swedish 1948 2550 12 5.7Eskimo 1954 348 1 4.4Canadian 1952 1071 7 5.6Yoruba 1961 245 2 6.8

the theory. This paper presents systematicdata from Eskimo and Yoruba groups, andinformation from several other culturalareas, which instead call the theory intoquestion.

Explicit labels for insanity exist in thesecultures. The labels refer to beliefs, feel-ings, and actions that are thought to ema-

nate from the mind or inner state of an in-dividual and to be essentially beyond hiscontrol; the afflicted persons seek the aid ofhealers; the afflictions bear strong resem-

blance to what we call schizophrenia. Ofsignal importance is the fact that the labelsof insanity refer not to single specific attri-butes but to a pattern of several interlinkedphenomena. Almost everywhere a patterncomposed of hallucinations, delusions, dis-orientations, and behavioral aberrationsappears to identify the idea of "losing one'smind," even though the content of thesemanifestations is colored by culturalbeliefs.The absence of a single label among Es-

kimos and Yorubas for some of the phe-nomena we call mental illness, such as

neuroses, does not mean that manifesta-tions of such phenomena are absent. Infact they form a major part of what theshamans and healers are called upon totreat. Eskimos and Yorubas react topeople they define as mentally ill with a

complex of responses involving first of allthe use of healing procedures but includingan ambivalent-appearing mixture of care

giving and social control. These reactionsare not greatly dissimilar from those thatoccur in Western society. Nor does theamount of mental illness seem to vary

greatly within or across the division ofWestern and non-Western areas. Patternssuch as schizophrenia, were, and nuthkavi-hak appear to be relatively rare in any one

human group but are broadly distributedamong human groups. Rather than beingsimply violations of the social norms ofparticular groups, as labeling theory sug-gests, symptoms of mental illness are man-

ifestations of a type of affliction shared byvirtually all mankind.

References and Notes

1. H. S. Becker, Outsiders (Free Press, New York,1963), reprinted in 1973 with a new chapter, "La-belling theory reconsidered."

2. T. S. Szasz, The Myth ofMental Illness: Founda-tions of a Theory of Personal Conduct (Hoeber-Harper, New York, 1961); R. Laing and A. Ester-son, Sanity, Madness, and the Family (BasicBooks, New York, 1964).

3. Notably in E. Lemert's distinction between prima-ry and secondary deviance, which appeared initial-ly in his book Social Pathology (McGraw-Hill,New York, 1951).

4. "In California, labeling theory itself contributed tothe formulation of the Lanterman-Petris-Shortact, a law which makes it difficult to commitpatients to mental hospitals, and still more diffi-cult to keep them there for long periods of time"(5, p. 256).

5. T. Scheff, Am. Sociol. Rev. 40, 252 (1975).6. D. Rosenhan, Science 179, 250 (1973).7. R. Benedict, J. Gen. Psychol. 10, 59 (1934).8. E. Schur, Labelling Deviant Behavior: Its Socio-

logical Implications (Harper & Row, New York,1971).

9. E. Lemert, Human Deviance, Social Problems andSocial Control (Prentice-Hall, Englewood Cliffs,N.J., 1967).

10. R. Scott, The Making of Blind Men (Russell SageFoundation, New York, 1969).

11. T. Scheff, Being Mentally Ill: A Sociological Theo-ry (Aldine, Chicago, 1966).

12. W. Gove, Am. Sociol. Rev. 35, 873 (1970).13. _ _ and P. Howell, ibid. 39, 86 (1974); W.

Gove, ibid. 40, 242 (1975); R. Chauncey, ibid., 248;T. Scheff (5, 14). In addition to the Rosenhanstudy (6), Scheff (14) evaluates the following six in-vestigations as strongly supporting labeling: J.Greenley, J. Health Soc. Behav. 13, 25 (1972); A.Linsky, Soc. Psychiatr. 5, 166 (1970); W. Rushing,Am. J. Sociol. 77, 511 (1971); M. Temerlin, J.Nerv. Ment. Dis. 147, 349 (1968); W. Wilde, J.Health Soc. Behav. 9, 215 (1968); D. Wenger andC. Fletcher, ibid. 10, 66 (1969). For several articlesunfavorable to labeling theory see W. Gove, Ed.,The Labelling of Deviance, Evaluating a Per-spective (Wiley, New York, 1975). Also not favor-able are R. Brown (15); N. Davis, Sociol. Q. 13,447 (1972); J. Gibbs, in Theoretical Perspectiveson Deviance, R. Scott and J. Douglas, Eds. (BasicBooks, New York, 1972), p. 39; S. Kety, Am. J.Psychiatr. 131, 957 (1974); R. Spitzer, J. Abnorm.Psychol. 84, 442 (1975); D. Ausubel, Am. Psychol.16,69 (1961).

14. T. Scheff, Am. Sociol. Rev. 39, 444 (1974).15. R. Brown, Am. Psychol. 28, 395 (1973).16. R. Benedict, Patterns of Culture (Houghton Mif-

flin, Boston, 1934); M. Mead, Sex and Temper-ament in Three Primitive Societies (Morrow, NewYork, 1935); Male and Female (Morrow, NewYork, 1949).

17. W. Caudill, in Transcultural Research in MentalHealth, W. Lebra, Ed. (Univ. Press of Hawaii,Honolulu, 1972), p. 25; R. Edgerton, in ChangingPerspectives in Mental Illness, S. Plog and R.Edgerton, Eds. (Holt, Rinehart & Winston, NewYork, 1969), p. 49; R. Levine, Culture, Personalityand Behavior (Aldine, Chicago, 1973); M. Field,Search for Security, An Ethnopsychiatric Study ofRural Ghana (Northwestern Univ. Press, Chicago,1960); A. Wallace, Culture and Personality (Ran-dom House, New York, 1970); B. Whiting and J.Whiting, Children ofSix Cultures (Harvard Univ.Press, Cambridge, Mass., 1975).

18. J. Honigmann, Personality in Culture (Harper &Row, New York, 1967).

19. R. Edgerton, The Cloak of Competence (Univ. ofCalifornia Press, Berkeley, 1967); J. Mercer, La-beling the Mentally Retarded (Univ. of CaliforniaPress, Berkeley, 1973).

20. R. Jessor, T. Graves, R. Hanson, S. Jessor, So-ciety, Personality and Deviant Behavior (Holt,Rinehart & Winston, New York, 1968); J. Levyand S. Kunitz, Indian Drinking (Wiley, New York,1974); H. Mulford, in The Mental Patient: Studiesin the Sociology of Deviance, S. Spitzer and N.Denzin, Eds. (McGraw-Hill, New York, 1968), p.155; D. Pittman and C. Snyder, Society, Cultureand Drinking Patterns (Wiley, New York, 1962).

21. S. Spitzer and N. Denzin, Eds., The MentalPatient: Studies in the Sociology of Deviance(McGraw-Hill, New York, 1968).

22. C. Fletcher and L. Reynolds, SocioL Focus 1, 9(1968).

23. E. Goffman, Asylums: Essays on the Social Situ-ation of Mental Patients and Other Inmates (Al-dine, Chicago, 1962), p. 132.

24. T. Sarbin, in Changing Perspectives in Mental Ill-ness, S. Plog and R. Edgerton, Eds. (Holt, Rine-hart & Winston, New York, 1969), pp. 11, 15, 19.

25. D. Mechanic, Ment. Hyg. 46, 68 (1962).

1027

26. G. Devereux, in Some Uses ofAnthropology: The-oretical and Applied, J. Casagrande and T. Glad-win, Eds. (Anthropological Society of Washing-ton, Washington, D.C., 1956), p. 23; A. Kroeber,The Nature of Culture (Univ. of Chicago Press,Chicago, 1952), pp. 310-319; R. Linton, Cultureand Mental Disorders (Thomas, Springfield, Ill.,1956), pp. 98, 118-124; J. Silverman, Am. Anthro-pol. 69 (No. 1), 21 (1967).

27. W. Rivers, Psychology and Ethnology (HarcourtBrace, New York, 1926), pp. 38-50.

28. B. Whorf, Language. Thought and Reality (Tech-nological Press of MIT, Cambridge, Mass., 1956).

29. J. Hughes [Murphy], thesis, Cornell University(1960). The recording of Eskimo words was con-ducted by Charles C. Hughes. The spelling givenhere follows the principles used in C. Hughes (withthe collaboration of J. Murphy), An Eskimo Vil-lage in the Modern World (Cornell Univ. Press,Ithaca, N.Y., 1960). The census of 1940 whichserved as a baseline was prepared by AlexanderLeighton and Dorothea Leighton. The extendedstatements by Eskimos and Yorubas which appearin quotation marks in the text are taken from myunpublished field notes, 1954-55, 1961, 1963. Mostof the Eskimo and Yoruba phrases are also takendirectly from these sources. In a few instances Ihave needed to paraphrase for intelligibility andtherefore I have not used quotation marks forphrases.

30. A. Leighton and J. Murphy, in Comparability inInternational Epidemiology, R. Acheson, Ed. (Mil-bank Memorial Fund, New York, 1965), p. 189.

31. A. Leighton, T. Lambo, C. Hughes, D. Leighton, J.Murphy, D. Macklin, Psychiatric Disorder amongthe Yoruba (Cornell Univ. Press, Ithaca, N.Y.,1963).

32. J. Douglas, The Social Meanings of Suicide(Princeton Univ. Press, Princeton, N.J., 1967).

33. Prince found that were was defined for him interms almost identical to those I present here; hestudied 46 were specialists (34, p. 84).

34. R. Prince, in Magic, Faith, and Healing, A. Kiev,Ed. (Free Press of Glencoe, New York, 1964).

35. J. Murphy, in Magic, Faith, and Healing, A. Kiev,Ed. (Free Press of Glencoe, New York, 1964), p.53.

36. M. Teicher, J. Ment. Sci. 100, 527 (1954).37. In looking through a magazine with me, an Eski-

mo pointed to a picture and said that it resembledthe shaman in seance; Fig. I is a photograph ofthat picture retouched to eliminate garmentswhich the Eskimo said were irrelevant to the sim-ilarity.

38. A. deReuck and R. Porter, Eds., Ciba FoundationSymposium. Transcultural Psychiatry (Churchill,London, 1965); A. Kiev, Transcultural Psychiatry(Free Press, New York, 1972).

39. S. Parker, Am. Anthropol. 62,603 (1960).40. Z. Gussow, in Psychoanalytic Study ofSociety, W.

Muensterberger, Ed. (International Univ. Press,New York, 1960), p. 218.

41. M. Teicher, Windigo Psychosis (American Eth-nological Society, Seattle, 1960).

42. Gussow (40) provides a description of pibloktoqbased on 14 recorded cases, mainly from explorersand ethnographers in the area from Greenland tothe west coast of Alaska during the first part ofthis century. Recently a serious attempt was madeto study pibloktoq properly and measure its preva-

lence. Ten cases were located from a population of11,000 Innuit Eskimos. These cases were found onfurther study to be exceedingly heterogeneous:"Several subjects had epilepsy-, several were diag-nosed as schizophrenic-, most had low normal se-rum calcium levels; one had hypomagnesemia andpossible alcoholism" [E. Foulks, The Arctic Hys-terias of the North Alaskan Eskimo (AmericanAnthropological Association, Washington, D.C.,1972), p. 117]. This information suggests that pib-loktoq is and may always have been a rare and ill-defined phenomenon. Regarding wittiko my assess-ment of the evidence is similar to Honigmann'swhen he says, "I can't find one [case] that satisfac-torily attests to someone being seriously obsessedby the idea of committing cannibalism" (18, p.401).

43. M. Beiser, J. Ravel, H. Collomb, C. Egelhoff, J.Nerv. Ment. Dis. 155, 77 (1972); M. Micklin, M.Durbin, C. Leon, Am. Ethnol. 1, 143 (1974); H. Ki-tano, in Changing Perspectives in Mental Illness,S. Plog and R. Edgerton, Eds. (Holt, Rinehart &Winston, New York, 1969), p. 256; R. Edgerton,Am. Anthropol. 68, 408 (1966); the following inMagic, Faith, and Healing, A. Kiev, Ed. (FreePress of Glencoe, New York, 1964): S. Fuchs, p.121; K. Schmidt, p. 139; M. Gelfand, p. 156; B.Kaplan and D. Johnson, p. 203; M. Whisson, p.283.

44. J. Murphy and A. Leighton, in Approaches toCross-Cultural Psychiatry, J. Murphy and A.Leighton, Eds. (Cornell Univ. Press, Ithaca, N.Y.,1965), p. 64.

45. A. Rose, in Human Behavior and Social Processes,A. Rose, Ed. (Houghton Mifflin, Boston, 1962), p.537.

46. Western society also lacked a comprehensive con-cept of neurosis prior to Freud's influence, but atthe present time neurotic patterns hold a firm posi-tion in the official classifications of Western psy-chiatry; see Diagnostic and Statistical Manual ofMental Disorders (American Psychiatric Associa-tion, Washington, D.C., 1968).

47. H. Selby, Zapotec Deviance, The Convergence ofFolk and Modern Sociology (Univ. of Texas Press,Austin, 1974).

48. Much of the support for labeling theory which Sel-by finds in his evidence stems from the followingstatement "To the villagers, witches have an ob-jective reality 'out there'. To me, they do not. I, thesociologist-anthropologist, do not believe thatthere are people in the world who have the capacityto float foreign objects through the air, insert theminto my body, and make me sick or kill me" (47,p. 13). He concludes, " We create the deviants; theyare products of our minds and our social process-es." It seems to me this is a mistaken conclusion. Iagree that the people who use witchcraft do not ac-tually kill their victims by their incantations, burn-ing effigies, boiling nail parings, and so on. Thequestion, however, is whether some people actuallycarry out these maliciously intended acts. Mywork with Eskimos and Yorubas suggests that theidea of witchcraft is widely available to thesegroups just as the idea of lethal weapons is to usand that a few people in such groups really do con-duct the rites that they believe will harm others(the artifacts of witchcraft attest to this), that theyare genuinely deviant in these practices, and thatthey are the brunt of strong disapproval because of

them. In this regard witchcraft involves real acts.It just happens that because these acts are by defi-nition secret they give rise to distortions, false ac-cusations, and misidentifications.

49. W. Bogoras, in The Jesup North Pacific Expedi-tion, F. Boas, Ed. (Memoir of the American Muse-um of Natural History, New York, 1904-1909), p.43.

50. T. Asuni, in Deuxi?me Colloque Africain de Psy-chiatrie (Association Universitaire pour le Devel-oppement de l'Enseignement et de la Culture enAfrique et a Madagascar, Paris, 1968), p. 115.

51. W. Bentz and J. Edgerton, Soc. Psychiatr. 6, 29(1971); H. Spiro, I. Siassi, G. Crocetti, ibid. 8, 32(1973).

52. F. Clements, Primitive Concepts ofDisease (Univ.of California Publications in American Arch-eology and Ethnology, Berkeley, 1932).

53. W. Dunham, Community and Schizophrenia, AnEpidemiological Analysis (Wayne State Univ.Press, Detroit, 1965), pp. 18, 19. Dunham indicatesthat 21 cases of schizophrenia were discovered inthe Essen-Maller study. I use 17 of these (those forwhom the author had high confidence that the pat-tern was schizophrenia). For comparability be-tween Tables I and 2, I recalculated the rate forthis one study. See also T. Lin, Psychiatry 16, 313(1953), for a similar use and interpretation of sev-eral of the studies cited here.

54. E. Essen-MOller, Individual Traits and Morbidityin a Swedish Rural Population (Ejnar Munks-gaard, Copenhagen, 1956). The rates for schizo-phrenia in the community and in the hospital werecalculated from information on pp. 85-86.

55. A. Leighton, My Name is Legion (Basic Books,New York, 1959); D. Leighton, J. Harding, D.Macklin, A. Macmillan, A. Leighton, The Charac-ter ofDanger (Basic Books, New York, 1963).

56. T. Lambo, in Magic, Faith, and Healing, A.. Kiev,Ed. (Free Press of Glencoe, New York, 1964), p.443; T. Asuni, Am. J. Psychiatry 124,763 (1967).

57. J. Murphy, in Transcultural Research in MentalHealth, W. Lebra, Ed. (Univ. Press of Hawaii,Honolulu, 1972), p. 213.

58. W. Weinberg, Arch. Rassen. Gesellschaftsbiol. 11,434 (1915). The Weinberg method of adjusting therate of mental illness for the probable age periodof susceptibility is useful when the age distribu-tions of the populations compared are different.Comparison of Western and non-Western popu-lations particularly need such adjustment. The ageof susceptibility for schizophrenia is assumed byWeinberg to be 16 to 40 years; I used 20 to 40years because that age breakdown is available inthe four studies compared.

59. The Eskimo and Yoruba studies which form thecore of this paper, and the Canadian study used forcomparison, have been carried on as part of theHarvard Program in Social Psychiatry directed byAlexander H. Leighton and supported by fundsfrom the Social Science Research Center of Cor-nell (for the Eskimo studies), the National Insti-tute of Mental Health, the Ministry of Health ofNigeria, and the Social Science Research Council(for the Yoruba studies), the Carnegie Corpo-ration of New York, the Department of NationalHealth and Welfare of Canada, the Department ofPublic Health of the Province of Nova Scotia, theFord Foundation, and the Milbank MemorialFund.

SCIENCE, VOL. 1911028