The establishment by treaty of pro-cedures for binding third-party settle-ment in cases where the ISA or thestate withholds permission for researchwithin their respective areas of controlis the position most favorable to basicresearch, and one that none of thecompeting interest groups at the 1973conference can effectively oppose. Hav-ing claimed that a legitimate distinctionexists between basic research and lim-ited exploration, the developed coun-tries cannot object to entrusting thisdetermination, when contested, to aneutral body of experts. Nor can de-veloping countries justify in the nameof the common heritage claims to theright to bar from their resource zonesresearch certified by an unbiased thirdparty to be in the common interest.Freedom for science is no longer a

universal right. But access to the oceansfor research "intended for the benefitof all mankind" is equally justified un-der the common heritage doctrine. Itwould be ironic if the new law of thesea that is being created to make this

doctrine a reality were to contain pro-visions which would impede the under-standing of the marine environment onwhich all nations, rich and poor alike,are going to depend increasingly in theyears ahead.

References and Notes

1. A. de Tocqueville, Democracy in America,P. Bradley, Ed. (Knopf, New York, 1963),vol. 1, p. 393.

2. Ocean Affairs Board of the National Academyof Sciences, "A proposed U.S. position onfreedom for science in the oceans" (un-puLblished, revised June 1972).

3. I make an obvious oversimplification intreating both the developed and the develop-ing countries as unified blocs. The varietyand complexity of the issues confronting the1973 Law of the Sea Conference, sponsoredby the United Nations, assures that votingwill not always be strictly along these lines.Science, however, is one of the issues inwhich the difference between developed anddeveloping countries is most apparent.

4. U.N. General Assembly Resolution 2750 (17December 1970), p. 2.

5. R. Friedheim, World Polit. 18, 25 (1965).6. As quoted by W. Marz, Biull. At. Sci. 24

(No. 9), 19 (1968).7. Representative of the developing country's

position in the debate over the ISA's juris-diction is a statement by the Peruvian dele-gate to the July 1971 meeting of the SeabedCommittee, who called for an authority em-powered to carry out a wide range of ac-tivities, including the coordination of scientific

researcli (see U.S. Department of State tele-gram from U.S. mission, Geneva, 241546Z,JuIly 1971).

8. Statement by the President on U.S. OceanPolicy, Oflice of the White House PressSecretary, 23 May 1970.

9. P. M. Fye, "Ocean policy and scientificfreedom," lecture given before the MarineTechnology Society, Washington, D.C., 11September 1972.

10. J. Bartlett, Bartlett's Familiar Quiotations, E.Beck, Ed. (Little, Brown, Boston, 1968), p.7(09.

l1. U.N. General Assembly Document A/Ac138/SR 54 (22 March 1971), p. 109; quotedby H. Franssen, in Freedom of Oceanic Re-search, W. S. Wooster, Ed. (Crane, Russak,New York, 1973), p. 158.

12. W. S. Wooster, "Costs and consequences ofrestrictions on research: Another view," dis-cussion paper presented at Scripps Institu-tion of Oceanography Center for MarineAffairs Workshop on Conditions for Free-domn of Oceanic Research, 24-26 April 1972.

13. K. Emery, Science 178, 298 (1972).14. P. M. Fye, personal communication.15. This proposal was made informally, and,

although it generated considerable comment,it was apparently never mentioned in officialrecords. I learned of it from A. E. Maxwell,provost of the Woods Hole OceanographicInstitution and a frequent delegate to thelOC.

16. This article would not have been possiblewithout the help of Robertson P. Dinsmore,Richard L. Haedrich, Herman Franssen,Maureen Franssen, Paul M. Fye, Kaleroy L.Hatzikon, and, above all, P. Sreenivasa Rao,who spent many hoturs introducing me to themysteries of international law. This article isWoods Hole Oceanographic Institution Con-tribution No. 3089.

Depressive disorders are perhaps themost distressful, and certainly amongthe most common, maladies that afflictmankind. Although man has knownand experienced melancholic statessince antiquity, it is only during thepast decade or so that we have begunto develop scientific insights into thebasic mechanisms involved.

Unfortunately, the literature on de-pressive disorders, like that on otherpsychiatric disorders, is composed ofisolated research reports, with few at-tempts at systematically integratingthem (I). Different schools of thought

20

utilize dissimilar dialects, thereby hin-dering communication, while ethicalconsiderations often preclude directtesting of hypotheses in human sub-jects. Studies are being carried out inan attempt to create experimental ani-mal models of depression-modelsthat would simulate the central fea-tures of human depressions. This ar-

ticle reviews these studies, as well as

other formulations, both clinical andmetapsychological, that derive fromdifferent frames of reference. We pre-sent evidence that depression in ani-mals is sufficiently analogous to hu-

man depression to offer insights intothe human condition. A comprehen-sive hypothesis of depression, incor-porating and synthesizing findings fromdifferent schools, is proposed.

Depression as a Final

Common Pathway

Several models of depression, reflect-ing diverse theoretical orientations,have been formulated.

1) The "aggression-turned-inward"model, originally proposed by Abra-ham and later elaborated by Freud, seesdepression as hostility turned inwardupon the loss of an ambivalently lovedperson (2). Although this is the mostwidely quoted psychological theory ofdepression, there is no systematic evi-dence to substantiate it (3). Indeed,this theory does not lend itself easilyto empirical verification because it isexpressed in metapsychological terms.

2) The "object loss" model, whichDr. Akiskal is assistant professor of psychiatry,

University of Tennessee College of Medicine, 42North Dunlap St., Memphis 38103, and researchpsychiatrist, Alcohol and Drug Research Center,Tennessee Psychiatric Hospital and Institute. Dr.McKinney is associate professor of psychiatry,University of Wisconsin School of Medicine, 1300University Avenue, Madison 53706, and researchpsychiatrist, University of Wisconsin PrimateLaboratory.

SCIENCE, VOL. 182

Depressive Disorders:Toward a Unified Hypothesis

Clinical, experimental, genetic, biochemical, and

neurophysiological data are integrated.

Hagop S. Akiskal and William T. McKinney, Jr.

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

also has its roots in psychoanalysis (4,5), views depression as a reaction tothe loss of a loved object. Althoughthe loss may involve symbolic posses-sions, such as one's values, status, andself-esteem (6), object loss generallyrefers to traumatic separation from aloved person-that is, the disruptionof an attachment bond.

3) The "reinforcement" model, whichutilizes behavioral concepts, postulatesthat depression is the name given tobehaviors that result from the loss ofmajor sources of reinforcement, fol-lowed by operant conditioning in theform of attention and sympathy (7).Depression is equated with chronicfrustration stemming from environ-mental stresses that are beyond thecoping ability of the individual, whoviews himself as being helpless andfinds relief in the rewards of the "sickrole."

4) Finally, the "biogenic amine"'model, which focuses on biochemicalderangements, hypothesizes a state ofthe central nervous system character-ized by depletion of biogenic amines(8). These neurotransmitters are con-centrated in the areas of the brain thatmediate arousal, sleep, appetite, sexdrive, and psychomotor activity (9),functions impaired in depression.We examine the interrelations among

these models and propose a biologicalfinal common pathway for depression.Specifically, we argue that depressivebehaviors must be understood as oc-curring on several levels simultaneouslyrather than as having a one-to-one, di-rect relationship with a single chemicalevent in the brain, and that a multi-plicity of genetic, developmental, phar-macological; and interyrsonal factorsconverge in the midbrain and lead toa reversible, functional derangementof the mechanisms of reinforcement.According to our hypothesis, then, ob-ject losses, interpersonally inducedstates of frustration and helplessness,and depletion of biogenic amines ul-timately result in impairment of theneurophysiological substrates of rein-forcement; this impairment is mani-fested behaviorally as depressive phe-nomena.

Heterogeneity in Clinical Depression

Defining and classifying depressivedisorders are crucial for research. AsKlerman suggests (10), some of theconfusion in research reports may de-rive from divergent concepts of what5 OCTOBER 1973

constitutes depressive behaviors andfrom patient populations that are diag-nostically heterogeneous.

Depression has been used to denotea variety of conditions, including (i) anormal mood state-for example, grief;(ii) a symptom synonymous with sad-ness that is seen in many psychiatricdisorders; and (iii) a syndrome char-acterized by psychomotor retardationor agitation, dejection, hopelessness,self-derogation, suicidal preoccupations,insomnia, loss of appetite (anoroia),and loss of libido. To avoid unneces-sary confusion, Whybrow proposes theuse of "depression" for mood states,while reserving "melancholia" for thesyndrome (11). A sustained state ofdeep dejection (melancholia) is a psy-chobiological final common pathway;once attained, the state becomes au-tonomous and assumes the dimensionsof illness-that is, morbidity, course,prognosis, and response to pharmaco-logical therapies. It should, therefore,be distinguished from the ubiquitous,transient, and minor changes in affec-tive responses which occur as part ofeveryday living.The American Psychiatric Associa-

tion, in the latest edition (1968) of itsdiagnostic and statistical manual (12),classifies depressions into two broadcategories: (i) those characterized byantecedent psychosocial conditions-"neurotic depression" and "psychoticdepressive reaction"-that are pre-sumably "reactive" to life events and,therefore, are more or less psycho-genic and (ii) "involutional melancholia"and "manic-depressive illness," where"the onset of the mood does not seemto be related directly to a precipitatinglife experience" (hence "endogenous,"or coming from within), with the impli-cation that they are biological in origin.The major problem with this system

of classification is that an etiologicalindex-presence or absence of precip-itating psychosocial events-is used indefining disorders whose etiology islargely unknown. Moreover, studieshave demonstrated (13) that the lackof psychosocial precipitants, thought tobe characteristic of the "endogenous"depressive, results from nonreporting.The severely depressed patient is toodisturbed to appraise fully the psycho-social context within which the illnessmanifests itself; upon clinical recovery,the frequency and type of stressfulevents, revealed by careful questioning,are no different from those in the "re-active" group. Depressive phenomenaare neither inherently psychosocial nor

biological (1, 14). As a final commonpathway, they are the culmination ofprocesses that can be described in manyframes of reference. Our hypothesis isan attempt to build conceptual bridgesbetween these various frames of refer-ence.

Robins and Guze propose a systemof classification that would excludeetiological considerations (15). Thelabel "secondary depression" refers todepressed patients with a history of adefinite psychiatric illness other thanaffective disorder. Currently, little isknown about secondary depression, be-cause there are no large-scale syste-matic studies of it (16). On the otherhand, "primary affective illness," whichis reserved for patients with no psy-chiatric disorder other than depressionand mania, has received great atten-tion from researchers during the pastdecade. It has been subdivided intotwo groups: unipolar affective disease,single or recurrent depressions; and bi-polar affective disease, characterized bymanic attacks in the subject or in hisclose biological relatives. Clinical (17),genetic (18), biochemical (19), phar-macological (20), and neurophysiologi-cal (21) differences have been de-scribed to further substantiate the uni-polar-bipolar dichotomy. For instance,bipolar depressives, as compared withunipolars, are characteristically retard-ed in psychomotor activity (22); havehigh genetic loading for affective ill-ness, suggesting dominant transmission(18); are more likely to have postpartumaffective episodes (23); will switch tomania upon treatment with sympatho-mimetic drugs (19); and their brainwaves (average evoked potentials) ex-hibit neurophysiological overreaction(augmentation responses) when pre-sented with visual stimuli-these re-

sponses can be corrected by adminis-tering lithium carbonate (21). The bi-polar type is clinically manifested inaffective episodes only (manias and de-pressions), and the vulnerability mightbe transmitted autosomally or by X-linkage (24). Unipolar depressions, onthe other hand, are clinically hetero-geneous and apparently are manifestedin forms other than depression, such asalcoholism, as suggested by the work ofWinokur et al. (25); the mode of in-heritance is uncertain, with some sug-gestion that it is polygenic (18).The heterogeneity of depressive dis-

orders encountered in clinical practicecan be understood as interactionsamong biological, psychological, andsociological factors (26). k view of

21

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Fig. 1. Hypothetical model of genetic-environmental interaction in depressive disorders.

the evidence for genetic heterogeneity,it is probable that there is more thanone biochemical form of depression.On the other hand, little attention hasbeen paid to the possibility that certainforms of depression result from inter-personal factors that secondarily in-duce biochemical changes in those areas

of the brain that modulate affect. De-spite the clinical and biological differ-ences described k\or the various cate-

gories of depression (secondary, bi-polar, unipolar), many symptoms thatconstitute the syndrome of depression(psychomotor and vegetative dysfunc-tion, dysphoria, hopelessness, suicidalpreoccupation) are common to the en-

tire group of depressive disorders. Thisuniformity of symptoms is consistentwith the hypothesis that a neurophysio-logical final common pathway underliesall types of depressive illness (see Fig.1).

Object Loss in Primates

The interaction of genetic and en-

vironmental factors in the genesis of

human depressions is complex. To

stuidy the psychosocial conditions that

permit the expression of depressive be-

haviors in a given individual, who mayor may not have a depressive geneticdiathesis, one should investigate each

of the psychosocial factors. For in-

stance, to state that separation from a

loved person might precipitate a de-

pressive condition is not very revealing.One must, for example, specify the ageat which separation occurs; the objectfrom whom separation takes place;the nature of the relationship between

the two before the separation; the man-

ner in which separation is achieved;

and, finally, the length of separationrequLired to cause depression. In other

22

words, one has to define the concept

of separation in a system where otherfactors are kept constant. Obviouslythis can be achieved most readily at

the animal level. Such experimentalparadigms would permit rigorous in-vestigation of the behavioral, develop-mental, and biochemical variables in-volved. One can do the same with theconcept of chronic frustration and help-lessness or any other psychosocial fac-tor thought to enter into the patho-genesis of depressive phenomena.The modeling of certain aspects of

the psychopathology of human de-pressions at the animal level had toawait the establishment of objectivecriteria for the evaluation of nonhumanpsychopathology (27, 28) and the pro-

duction of stable behavioral syndromes.Otherwise, the situation would havebeen similar to clinical psychiatric re-

search, where confusion in the classi-fication of disorders has often pre-

vented generalization of findings fromone study to another.The nature and variety of reactions

to separation have been studied exten-sively in nonhuman primates becausethese animals form very strong at-tachment bonds. These experimentswere stimulated by clinical researchundertaken by child psychoanalysts.

Spitz (4) reported a deprivationalreaction in human infants separatedfrom their mothers in the second halfof their first year. The reaction was

characterized by: (i) apprehension and

crying; (ii) withdrawal; (iii) gross re-

tardation of development, retardationof reaction to stimuli, slowness of move-

ment, dejection, and stupor; and (iv)anorexia, weight loss, and insomnia. This

syndrome, called anaclitic depression,could, unless reversed by providing a

substitute mother, result in death

from inanition and superimposed infec-

tion. It is interesting that only 15 per-cent of the infants suffered anacliticdepression. According to Spitz, thoseinfants who had enjoyed a gratifyingrelationship with their mothers werethe ones most susceptible to the traumaof separation. Yet one should also con-sider the possible interaction of geneticvulnerability with loss of the mother ata critical age.A similar reaction was described in

older children by Robertson and Bowl-by (5). It had three distinct phases: a,"protest" phase, during which the childexhibited restlessness and tearfulness insearch for the mother; a "despair"phase, in which the behavior of thechild was characterized by apatheticwithdrawal; and a final "detachment"phase, in which the child rejected themother upon reunion.

These investigations stimulated re-search in infrahuman primates, withthe purpose of creating animal modelsof anaclitic depression. Starting in thelate 1950's, many experiments were de-signed at the Wisconsin Primate Lab-oratory to study the differential effectsof maternal separation and social iso-lation on rhesus monkeys (Macacainidlatta) at different ages (29). Separa-tion from the mother during infancyprovided the best parallel to humananaclitic depression (30). Monkeys (5to 7 months old) that were separatedfrom their mothers but still living withtheir peers in their original playpen,went through a predictable course ofagitation in search for their mothersto almost total lack of interaction withtheir peers and their environment. Thiswas a rather stable behavioral syn-drome, reproduced and amplified byother investigators (31). These stagesof separation are illustrated in Fig. 2.

In one of the Kaufman-Rosenblumexperiments (32), four pigtail (Macacaneinestrina) infants (4 to 5 montbsold) that had been reared in a labora-tory pen in a group living situation withtheir feral mothers were separated fromtheir mothers for a period of 4 weeks.The mothers were removed from theoriginal pen, leaving the four infantstogether. The first 24 to 36 hours weremarked by "acute distress" (protestphase): loud screams, searching headmovements, pacing, and restlessness.During the next 5 to 6 days, the in-fants were described as "depressed"(despair phase): they "sat hunchedover, almost rolled into a ball," withthe head down between the legs, andindulged in self-mouthing; their be-havior was generally withdrawn, with

SCIENCE, VOL. 182

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

only occasional reference to inanimateobjects and almost no interaction withpeers.

Anaclitic depression is complicatedby the fact that, in addition to the dis-ruption of a strong attachment bond,the loss of the relationship with themother could markedly interfere withthe infant's adaptive responses and, ulti-mately, survival. Studies of infants sep-arated from their peers (33) circum-vent this problem, because separationfrom peers interferes almost exclusivelywith the attachment bond. The experi-mental paradigm consisted of eithera single, protracted separation or mul-tiple, short separations of a groupof four infant macaques that had beenreared together since birth. During sep-aration, they went through a protestphase, with excessive vocalization andrandom locomotion, followed by adespair phase, with marked increasein self-clasping and huddling andmarked decrease in locomotion andexploration. One can conclude that thedisruption of an attachment bond,whether infant-mother or infant-infant,is a powerful inducer of psychopathol-ogy of the "depressive" type.The age at which separation takes

place and prior experience with sep-aration seem to be important factors indetermining the form that reactions toseparation take. Juvenile monkeys (3to 4 years old) ordinarily react to sep-aration only with behavior character-istic of the protest phase (34). Hence,great caution should be exercised inviewing anaclitic depression as theprototype for adult depressions. On theother hand, a Wisconsin study (35)found that juvenile monkeys that hadundergone traumatic separation duringinfancy responded to separation in amanner reminiscent of the despairphase. This suggests that early breaksin attachment bonds could predisposeone to adult depression.

Pharmacological "Depression"

Chemically induced psychopatholog-ical states in primates provide furtherparallels with human depression. Con-sistent with the biogenic amine hy-pothesis is the fact that reserpine, adepletor of biogenic amines, precipi-tates depression in 15 to 20 percent ofhypertensive human patients whengiven in doses of more than 0.5 milli-gram per day (36). In a third of thesepatients, the depression is seriousenough to require psychiatric hospital-5 OCTOBER 1973

Protest stage

Despair stageFig. 2. Illustration of typical protest anddespair stages following separation fromthe mother.

ization and electroconvulsive therapy.For this reason, the reactions of pri-mates to reserpine have been tested.Rhesus monkeys fed reserpine for 81days exhibited significant decreases inlocomotion and visual exploration andan increase in huddling behavior (37).The resemblance to the despair phaseof both human and primate infants isapparent (Fig. 3). It would be inter-esting to compare the effect of reser-pine on normally reared adult monkeyswith its effect on adult monkeys witha history of traumatic separation. Thiswould more closely parallel the reser-pine story in man, since severe depres-sion is induced most readily in personswith a history of depressive illrness ora family history of such illness (36).

Alpha-methylparatyrosine (AMPT),

Fig. 3. Typical huddling posture in rhesusmonkey following administration of reser-pine.

a selective depletor of dopamine andnorepinephrine (38), was administeredto adult stump-tail monkeys (Macacaspeciosa) by Redmond et al. at theIllinois State Psychiatric Institute (39).The animals (which exhibited no signsof toxicity at the end of the study)displayed marked changes in behavior5 to 6 weeks after treatment withAMPT was begun: (i) decreased totalsocial interactions; (ii) decreased socialinitiative; (iii) changes in pos-ture andfacial expression, suggesting withdrawaland lack of concern with the environ-ment; and (iv) retarded motor activ-ity, without evidence of extrapyramidalreactions. The Wisconsin group foundsimilar results with AMPT (40). Boththe Illinois (41) and Wisconsin (40)investigators, however, found no ap-preciable effects on the behavior ofmonkeys treated with parachlorphenyl-alanine, the selective depletor of sero-tonin (42). Monkeys treated withAMPT, despite their "retarded" motorappearance and behavior, exhibitedelectroencephalographic evidence ofarousal. This is consistent with studiesin man; contrary to classical notions,evidence for hyperarousal has beenfound in depressed individuals (43).When injected into the central ner-

vous system, 6-hydroxydopamine (6-OHDA) causes permanent destructionof central catecholamine fibers, whileleaving the peripheral sympathetic sys-tem, and both the central and periph-eral serotonergic systems, intact (44).Stein and Wise (45, 46) have hypothe-sized that 6-OHDA is somehow in-volved in the etiology of schizophrenia(and perhaps also of manic-depressiveillness). Whether or not this is found tobe true, it is now established that6-OHDA given intraventricularly torats and rhesus monkeys produces sig-nificant changes in behavior. In theadult rat, there is a temporary diminu-tion in activity and in eating and drink-ing, a reduction in the rate of self-stimulation in the brain, and failure tolearn to avoid unpleasant events inactive avoidance tasks (47). In rhesusmonkeys, there is a dramatic decreasein locomotion and social interactionand increased passivity (48) (see Fig.4). These changes in behavior occurafter each injection of 6-OHDA; theyare temporary, however, despite a per-manent depletion of norepinephrine inthe brain. This phenomenon requiresfurther investigation, but is generallyconsistent with the hypothesis pre-sented here. The data from studies ofmonkeys support the notion that de-

23

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

pression is a final common pathway,rather than a specific chemical event,and that it represents the response ofthe central nervous system to multiplestresses that could be described in inter-personal or biochemical language.

Helplessness and Depression

Is it possible to induce helplessnessin animals-that is, a state in whichthe animal can no longer cope with thefrustrations in its environment andsimply gives up? For many clinicians,hopelessness and helplessness representthe central feature of human depres-sions (49). For instance, Beck and hisassociates (50) found that "negativecognitive set," the perception of one-self as being helpless and hopeless andhaving no control over one's fate, cor-relates best with the depth of depres-sion and suicidal behavior.

"Learned helplessness" is an opera-tional construct which Seligman et al.(51) use to refer to a stable behavioralpattern characterized by failure to ini-tiate responses to escape traumaticevents and failure to learn that one'sown responses could be instrumentalin terminating noxious stimuli. Dogswere subjected to repeated, inescapableelectric shock while strapped in aPavlovian harness; when these dogslater received electric shock in a shuttle-box, they failed to cross the barrier,thereby "passively" accepting the highlytraumatic shock experience. This be-havior was in remarkable contrast tothat of dogs which had not been previ-ously exposed to inescapable shock;during shuttlebox training, they im-mediately learned that their response-to jump the barrier-ended the shock.Learned helplessness was reversed byforcibly dragging the dog to the otherside of the shuttlebox; apparently thedog learned that its own responsescould bring relief from noxious stim-ulation.

In Seligman's view (28), learnedhelplessness parallels clinical depres-sions in which the individual loses con-

trol over the reinforcers in the envi-ronment. Negative expectations aboutthe effectiveness of one's efforts inbringing the environment under one'scontrol (hopelessness, helplessness, pow-ecrlessness) lead to passivity and dimin-ished initiation of responses (retarda-tion of psychomotor activity andthought processes, seen clinically).Such clinical states of helplessness

and passivity do not seem to indicate

24

Fig. 4. Acute behavioral effects followingintraventricular administration of 6-hy-droxydopamine.

actual deficiencies in the behavioralrepertoire of the patients. Some studieshave found that depressed patients ac-

tively attempt to manipulate their en-

vironments (52); others have found no

evidence of objective deficiencies inactual performance when patients were

coerced into engaging in the task athand (53). One is led to conclude thatdepressed patients have an underlyingdisorder of motivation and that theirability to derive reinforcement fromtheir environments is impairedWhat interpersonal situafions lead to

such maladaptive patterns of behavior?For those, in the Wolpian tradition

(54), chronic aversive stimulation-that is, chronic failure in one's attemptsto have the upper hand in interpersonalrelationships-culminates in chronicanxiety and inability to reduce anxietyby means of one's usual behavioralrepertoire. The clinical picture of suchmaladaptive behavior, characterized bypassivity and hopelessness, is somewhatsimilar to that of Seligman's dogs. It isclaimed that systematic desensitization,aimed at reducing the anxiety, or as-

sertive training, designed to help theindividual gain control over interper-sonal contingencies, are successful intreating such "neurotic depressions."

Lewinshon et al. (55), who utilize a

Skinnerian frame of reference, statethat a low rate of positive reinforce-ment is the antecedent of depression,which is further maintained by posi-tive reinforcement in the form of sym-pathy and attention. Lack of "social

skill," defined as behaviors that are

seldom positively reinforced by others,is viewed as the central behavioraldeficit. This concept is useful becauseit emphasizes the fact that it is not

passivity per se which characterizes the

depressive, but behaviors that do notreinforce him, no matter how "active"he is. Observations of both the patientand his family at home provide a pow-erful tool in manipulating interpersonalcontingencies that elicit or maintain de-pressive behaviors. Indeed, managinginterpersonal contingencies has provedsuccessful in treating depressed individ-uals (56). Positive reinforcement isprovided for active behaviors that takethe place of the lost source of rein-forcement; such reinforcement, coupledwith selective inattention to depressivebehaviors, leads to their extinction.

In summary, chronic aversive stim-ulation, loss of reinforcement, and lossof control over reinforcement are over-lapping concepts that describe a stateof hopelessness and helplessness deriv-ing from interpersonal relations. Avariety of techniques, deriving fromboth classical and instrumental condi-tioning, could be utilized to alleviatesuch depressive s-tates, but apparentlythese techniques are useful in themilder, so-called neurotic, depressionsand that more severe depressionsusually require antidepressant drugs orelectroconvulsive therapy (54). Thereare two possible explanations for thisphenomenon: (i) no matter what inter-personal factors elicit or maintain de-pressive behaviors, once these behaviorsassume severe proportions they becomebiologically autonomous-the stage ofmelancholia (11)-and, consequently,require somatic therapies; (ii) severedepressions have underlying biochemi-cal predispositions and, therefore,would not respond to any appreciabledegree to verbal therapy.

Biogenic Amines and Depression

Disturbances in both classes of bio-genic amines, the catecholamines (dopa-mine and noradrenaline) and the indole-amines (serotonin and tryptamine),have been hypothesized to cause af-fective disorders (8). The catechol-amine hypothesis, the first biogenicamine hypothesis to be formulated ex-

plicitly, states that depression is asso-ciated with a deficiency of catechol-amines, particularly norepinephrine, inthe central nervous system, and maniawith an excess of catecholamine. Thepossible role of impaired biogenicamine function in these disorders wasfirst inferred from pharmacological evi-dence (57). It was known, for instance,that a syndrome resembling naturallyoccurring depressions would develop in

SCINCE, VOL 182

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

hypertensive patients being given largedoses of reserpine and alpha-methyl-dopa over a period of, on the average,3 months. Both drugs are biogenicamine depletors. It now appears, how-ever, that reserpine is more of a chem-ical trigger, which upsets a vulnerablebiochemical-neurophysiological system,than a primary cause of depression (36).As for the pharmacological therapieseffective in depression, they are knownto raise the level of free amnines in thebrain (57). Tricyclic antidepressantssuch as imipramine and amitriptylineprevent the reuptake of biogenic aminesreleased from the presynaptic stores,thereby causing a net increase in thesetransmitters at the postsynaptic junc-tion. Monoamine oxidase inhibitors suchas tranylcypromine and phenelzine pre-vent the oxidative deamination of theseamines at the presynaptic stores, againincreasing their availability at the re-ceptor sites on the postsynaptic mem-brane.

Because the effects of reserpine andthe antidepressant drugs on biogenicamine metabolism, storage, and releaseare nonspecific, the pharmacologicalevidence thus far reviewed does notdistinguish between the role of indole-amines and that of catecholamines.Depressed patients have been given themetabolic precursors of these aminesin an attempt to discriminate betweentheir respective effects. A collaborationbetween U.S. and British researchers(58) has strengthened the earlier Britishclaims that L-tryptophan, the aminoacid precursor of serotonin, has anti-depressant properties, especially whencoupled with a monoamine oxidase in-hibitor (59). The catecholamine pre-cursor, L-dopa, on the other hand, hasbeen a failure in therapy, except for asmall subgroup of retarded depressives(60). The precursor loading strategy,therefore, lends support to the hypoth-esis that a deficiency of indoleaminesin the central nervous system plays animportant etiological role in the patho-genesis of depressive disorders (61). Italso suggests that retarded psychomotoractivity, a behavioral correlate of cate-cholaminergic deficiency in the centralnervous system, could be of etiologicalsignificance in some depressed patients(62). Yet, if optimal states of moodand psychomotor functioning dependon the harmonious functioning of sero-tonergic and catecholaminergic sys-tems in the diencephalon, then thenegative results with L-dopa could per-haps be explained by the decreasedserotonin synthesis that has been ob-5 OCTOBER 1973

served in depressed patients receivingL-dopa (63).

Definitive evidence for the biogenicamine hypotheses has to await thedemonstration of significant alterationsof monoamines in the brains of de-pressed patients. Although 5-hydroxy-indoleacetic acid (5-HIAA), the majormetabolite of serotonin, and 5-hydroxy-tryptamine have been found in lowconcentrations in the hindbrain of de-pressive suicides (64), one must re-member that postmortem neurochemi-cal determinations are beset with meth-odological problems. However, analysisof the cerebrospinal fluid from de-pressed patients-and, interestingly,from manic patients-has revealed lowconcentrations of 5-HIAA (65), which,at least in depression, persist withclinical recovery (61). Since concen-trations of 5-HIAA in cerebrospinalfluid are affected by peripheral (gastro-intestinal) sources of indoleamines, theprobenecid technique of preventing theactive transport of organic acids acrossthe blood-brain barrier has been intro-duced into clinical research; the resultsobtained with the probenecid modifi-cation, however, have been largely in-conclusive, with some suggestion thatboth dopamine and serotonin metabo-lites are lowered in concentration (66).As for noradrenaline, about 30 per-

cent of the 3-methoxy-4-hydroxyphenyl-ethylene glycol (MHPG) in the urineis thought to come from the metabolicdegradation of noradrenaline in thecentral nervous system (67). This me-tabolite has been found in subnormalconcentrations in depressed patients, re-turning to normal after treatment withimipramine (68). However, it is notclear at present whether low concen-trations of MHPG, and low concentra-tions of metabolites of other biogenicamines correlate best with retardedpsychomotor function or with de-pressed mood (69).

In summary, the available data onbiogenic amines and depressive disor-ders do not distinguish between one bio-genic amine and another. Although it ispossible that distinct groups of de-pressed patients suffer depletion of aparticular biogenic amine, a depletionmodel based on one biogenic amine isprobably an oversimplification. Oneshould perhaps consider the possibilityof inefficient interactions of theseamines or substitution by faulty neuro-transmitters (70). The failure of indole-amines to return to normal after thepatient's clinical recovery from de-pression has led to the hypothesis that

serotonergic deficiency in the centralnervous system determines the patient'spredisposition to depressive (and manic)illness; while the depressive episode istriggered by catecholamine depletion(and mania is triggered by an increasein catecholamines). In this view, de-pression and mania are on a con-tinuum, sharing the same basic dys-function, rather than being polar op-posites, with mania representing a moresevere deviation from normal mood (71).Bunney, Goodwin, and Murphy havesuggested that the available data pointto the role of catecholamine excess inexacerbations of manic episodes, butthat the underlying dysfunction in bothmania and depressions of the bipolarvariety probably involves a basic in-stability of the presynaptic neuronalmembrane; such dysfunction is per-haps partially correctable by imipramineor lithium carbonate, or both (72).Another hypothesis, postulated by Ja-nowsky, El-Yousef, and Davis, attrib-utes depression to a shift in the balanceof norepinephrine and acetylcholine inthe diencephalon toward the choliner-gic side (73). Finally, the work ofPrange et al. suggests impaired sensi-tivity of the postsynaptic monoamin-ergic receptor (74), a condition thatcan be partially corrected by admin-istering thyroid hormone or thyroid-stimulating hormone on top of the tri-cyclic antidepressant regimen (75).This is an important consideration, sincesynaptic transmission is determinedjointly by presynaptic changes in bio-genic amines and by postsynaptic re-ceptor sensitivity. With impaired re-ceptor sensitivity present, it is possiblethat depression could exist with normalor high levels of catecholamines.

Attachment, Reinforcement,and Biogenic Amines

The original psychoanalytic model,paraphrased by Spitz (4) and byRobertson and Bowlby (5) as a reac-tion to the loss of the love object, canalso be conceived of in terms of dis-ruption of an attachment bond, as sug-gested by primatologists (76).

Learning theorists like Dollard andMiller (77) regard the love of the in-fant for its mother as a learned "sec-ondary dependency drive" derivedfrom the reinforcement of nursing (78).However, the Harlows have demon-strated that the affection of the infantfor its mother is independent of feed-ing and that attachment is a powerful

25

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

INTERPERSONAL DEVELOPMENTAL

ABNORMALGENE§)

OBJECT LOSSDURING CHILDHOOD

IREVERSIBLE FUNCTIONAL DERANGEMENT OF THE DIENCEPHALIC MECHANISMS OF REINFORCEMENT

| DEPRESSIVE DISORDERS IFig. 5. A multidisciplinary model of the pathogenesis of depressive disorders.

"primary drive" in its own right, jusWas feeding and sex are (79). The at-tachment of one human being to an-

other, be that infant-mother or peer-

peer, is a powerful reinforcer, and itsloss can lead to serious psychopathol-ogy. This fact gives psychoanalytic,ethological, and behavioral theories a

common denominator: depression isassociated with the loss of significantreinforcers. Although the behavioralformulation is a broader one, it is a

common clinical observation that hu-man depression is usually a reaction towithdrawal of reinforcement from sig-nificant others. In primate species, in-cluding man, object losses in the formof disrupted attachment bonds are

probably the most traumatic examplesof loss of reinforcement. Seligman'sformulation is even broader, because it

views loss of reinforcement as a spe-

cial instance of loss of control over re-

inforcement. In conclusion, loss of rein-forcement, however achieved, seems to

be antecedent to a perception of oneself

as having lost one's ability to exercisefuture control over such reinforcers.

Is it possible to reconcile this be-

havioral-ethological formulation with

the biochemical point of view? Stein's

work (80) links the pharmacology of

depression with the neurophysiologicalsubstrates of motivation and reward.

Thus amphetamines, euphoriant agents,when injected through permanently im-

planted electrodes into the "reward cen-

ters," increase self-stimulation. Imipra-mine, an antidepressant, augments the

action of amphetamines. Reserpine and

26

AMPT, which are known to precipi-tate a depression-like syndrome, de-crease the action of amphetamines.

Extensive studies of the neurophysi-ological substrates of reinforcementhave elucidated the biology of rewardand punishment (80, 81). The medialforebrain bundle (MFB), the anatomi-cal substrate for the "reward system,"originates in the locus coerulus and theadjacent reticular formation and formsnoradrenergic synapses in the lateralhypothalamus, at higher levels in thelimbic system, and in the frontal cor-

tex. The periventricular system (PVS),the anatomical substrate for the "pun-

ishment system," arising in the dorso-medial region of the midbrain, formscholinergic synapses in the medialhypothalamus and elsewhere in the lim-bic system. It appears that noradrena-line exercises inhibitory control over

behavior-suppressant cell groups in theforebrain, while acetylcholine facili-tates the activity of these nuclei. Itshould be noted that stimulation of

certain portions of the noradrenergicsystem enhances sexual activity; where-

as lesions in this system (for example,in the lateral hypothalamus) produceanorexia. Stimulation of the PVS re-

sults in behavioral reactions usually as-

sociated with pain, such as jumping,biting, and screaming. Finally, simula-

tion along the punishment and reward

pathways is analogous to the action of

primary reinforcers, in that neutralstimuli, presented just before stimula-tion, acquire secondary reinforcingproperties.

Thus "one may think of the norad-renergic medial forebrain bundle fibersas part of a behavior-facilitating or

'go' mechanism [that] . . . initiates fa-cilitatory feedback [and] . . . increasesthe probability that the behavior willrun off to completion" (46, p. 348).The MFB contains serotonergic fiberstoo (82).

Toward a Unified Hypothesis

From the foregoing information one

can predict that impairment of biogenicamine function would result in dimin-ished initiation of responses, decreasedactivity, and disturbances in appetite,sex drive (9), and sleep (83). It alsoprovides a conceptual bridge betweenthe behavioral and the biological mod-els of depression. Loss of reinforcementinfluences the behavior of the orga-

nism through its effects on the dien-cephalic mechanisms of reward andpunishment. For instance, convention-al reinforcers have no effect on be-havior when the anatomical or chemi-cal integrity of the self-stimulation sys-

tem of the diencephalon is disrupted(84).

If depressive behaviors elicited by a

variety of mechanisms have as theirneurophysiological final common path-way a reversible, functional derange-ment in the diencephalic mechanismsof reinforcement, as we have hypoth-esized, then chemical, genetic, develop-mental, and interpersonal-experientialfactors should all impinge on the dien-cephalic centers of reward, as shown inFig. 5.

1) Chemical. Certain drugs or meta-bolic disorders can suppress or interferewith the neurotransmitters of this sys-

tem-for example, reserpine and alpha-methyldopa in man, reserpine and al-pha-methylparatyrosine ii7tiifrahumanprimates; and "borderline hypothy-roidism" which could chemically in-terfere with the sensitivity of themonoaminergic receptor.

2) Genetic. These factors must actby means of their effects on the chem-istry of reinforcement, because chem-istry is the only language into which

genes translate their message in con-

trolling organismic function. There is

substantial evidence for the contribu-tion of heredity in some, if not all,forms of primary affective illnesses (18).However, the specific enzymatic defects

that lead to the chemical derangementsare unknown. If instability of the neu-

ronal membrane is the underlying dys-SCIENCE, VOL. 182

CHEMICAL GENETIC

Hypothesized Biochemical Intermediates1. Impaired Neuronal ATPase System2. Decrease in the Functional Level of Noreplneophrlne3. Decrease in the Functional Level of Serotonin4. Imbalance in the Serotonergic- Norsdrenerglc Systems5. Faulty Neurotransmitters6. Cholinergic Dominance7. Inpaired Sensitivity of the Monoominergtc Poskynapttc Receptor

-

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

function in affective illness (72), thenone possible enzymatic defect might bein the mechanism that controls sodiumextrusion from the neuron-that is, *aderangement in neuronal membraneadenosine triphosphatase. The workof Coppen and his associates providesevidence that there may be, in depres-sion, an accumulation of intraneuronalsodium (85). It is also known that anincrease in intraneuronal sodium re-sults in extraneuronal leakage of bio-genic amines (86). This hypotheticalchain of events can be summarized asfollows: Abnormal gene -* (periodic)adenosine triphosphatase deficiencyincreased intraneuronal sodium -> bio-genic amine depletion -- derangementof diencephalic mechanisms of rein-forcement -- failure to respond to re-inforcement -- depressive behaviors(87). As indicated earlier, there aremost likely several distinct genetic de-fects that could lead to various pat-terns of impairment in the functionallevel or activity of biogenic amines inthe diencephalon. Relative insensitivityof the monoaminergic receptors (74),which could be caused by a geneticallytransmitted, defective structural lipo-protein of the postsynaptic membrane,is another possibility. According to thescheme presented here, and irrespectiveof the individual lesions, they would allresult in a decrease in the functionalcapacity of the reward system; or a pre-dominance of the punishment system-that is, cholinergic dominance (73, 80,82), to phrase it in neurophysiologicalterms.

3) Interpersonal. The occurrence ofanaclitic depression in human and pri-mate infants deprived of or separatedfrom their mothers can be readily un-derstood if one remembers the workof the Harlows, which showed at-tachment behavior to be a primarydrive (79). Its neurophysiological sub-strate probably resides in the dienceph-alon, as is true of other primary drives.As noted earlier, depressions in humanadults are often preceded by the with-drawal of reinforcement from signifi-cant others, usually in the form ofbreaks in attachment bonds. Such astate of affairs could conceivably leadto (reversible) derangements in the di-encephalon, which would in turn leadto the dominance of the periventricularpunishment system.

But is there any evidence that experi-ential factors lead to derangements ofthe biological substrates of reinforce-ment? Weiss et al. (88) demonstratedthat norepinephrine is depleted in the5 OCIOBER 1973

brain of animals exposed to aversivestimuli beyond their coping ability;these animals also suffered anorexia andweight loss. On the other hand, thoseanimals that could cope with or avoidthe aversive stimulus did not exhibitsuch chemical and somatic derange-ments. To extrapolate to human beings,interpersonally induced states of help-lessness, whether from chronic aver-sive stimulation or loss of control overreinforcement contingencies, could re-sult in alterations of biogenic amines.

4) Developmental. Psychoanalystsclaim that bereavement in childhoodpredisposes one to depression in adult-hood. Epidemiological evidence for thishypothesis has been generally negativeor inconclusive (89). It is possible, how-ever, that if the bereavement occursat certain critical ages it can predis-pose to adult depression. Furthermore,a good parental substitute might pre-vent the adverse effects of childhoodbereavement. On the other hand, bothchildhood bereavement and adult de-pression might be caused by the samefactor (for example, genetic vulner-ability to depression), since it has beendemonstrated that mortality in bipolarillness, both from suicide and othercauses, is very high (18). Therefore,one would expect patients from familieswith bipolar illness to suffer bereave-ment during childhood and depressionduring adult life. Future epidemiologi-cal work should control for these vari-ables. Primate models of depressionmay help clarify this Issue. In a recentstudy (35), 2-year-old monkeys wereshown to respond differently to separa-tion, depending on whether or not theyhad undergone traumatic separationduring infancy; those with a history ofseparation responded with increasedself-directed behaviors and social with-drawal, as compared to control subjects.As to the way in which childhood be-reavement operates, in terms of themodel presented here, one can hypoth-esize that early breaks in attachmentbonds stunt the maturational poten-tial of the brain mechanisms that sub-serve reinforcement, making them morevulnerable to adult disappointment andfrustration-that is, the individual hasa "fragile reward system," one whichcan be easily upset.

5) Genetic-interpersonal -Probablygenetic predisposition in the form of a"labile diencephalic reinforcement sys-tem" (neuronal membrane instability?impaired sensitivity of monoaminergicreceptors?) renders 10 to 15 percentof the population (18) extremely vul-

nerable to the effects of interpersonalconflicts or events that result in a de-crease in positive reinforcement. Thatreinforcing events from the environ-ment converge and interact with the re-ward centers of the hypothalamus hasbeen demonstrated in infrahuman ani-mals (84); and there is no reason tobelieve that man has two separate sys-tems of reinforcement, or that inter-personally defined reinforcement by-passes the diencephalon.One might raise the question of why

genetic vulnerability to depression mostcommonly manifests itself in middleage and later. Two possibilities suggestthemselves: (i) the stresses of lateadulthood are such that they result insubstantial decrease in positive rein-forcement or loss of control over one'sdestiny-for example, physical decline,chronic illnesses, retirement, financialtroubles, loss of children through mar-riage, and loss of friends through death;and (ii) if an optimally functioning dien-cephalic reward system depends on ad-equate levels of monamine neurotrans-mitters, then a decrease of such amineswith age-which would be expectedfrom the finding that concentrations ofthe enzyme monoamine oxidase in thebrain increase with age (90)-mightprovide another answer to the in-creased vulnerability of the elderly todepression.Melancholia (11), then, can be looked

upon as a psychobiological state that isthe final common pathway of processesinvolving interpersonally induced statesof mind in which the individual seeshimself as losing control over his fate(hopelessness); increased psychic tur-moil; increased neuronal excitabilityand hyperarousal; disturbance of ge-netically vulnerable neuronal circuits inthe diencephalon; depletion of biogenicamines; impairment of the neurophysi-ological substrates of reinforcement;further decrements in coping mecha-nisms; and a vicious cycle of more hope-lessness, psychic turmoil, and hyper-arousal.

Summary

Our scientific understanding of psy-chiatric syndromes, including the phe-nomena of depression, has been ham-pered because of: (i) the use ofmetapsychological concepts that aredifficult to test; (ii) methodological andlinguistic barriers that prevent commu-nication among psychoanalysts, behav-iorists, experimental psychologists, and

27

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

psychiatrists; and (iii) the reluctanceof psychiatrists to accept animal mod-els as possible approximations of cer-tain aspects of human psychopathology.We have attempted to demonstrate

that the animal models simulate someof the central features of clinical de-pression (for example, helplessness andobject loss), thereby allowing one torigorously investigate them from de-velopmental, behavioral, and biochem-ical perspectives.The object loss model, as a concrete

version of a metapsychological-psycho-analytic concept, has enabled prima-tologists to study the disruption of anattachment bond. The behavioral modelaccommodates this concept to a broad-er generalization: loss of reinforcementor loss of control over reinforcement.We have reviewed the evidence thatthese processes involve the dience-phalic centers of reward or reinforce-ment, thereby permitting integration ofthe psychoanalytical and behavioralformulations with the biochemical hy-potheses. Also, we have presented datastrongly suggesting that the breaking ofan attachment bond in the primate rep-resents significant loss of reinforcementthat induces helplessness and disruptsmotivated behavior.

Finally, we have argued that the de-pressive syndrome could be caused byinteractions of genetic, chemical, de-velopmental, and interpersonal factors,all of whicir impinae on the dience-phalic centers of reinforcement.

References and Notes

1. H. Akiskal and W. McKinney, Arch. Gen.Psychiat. 28, 367 (1973).

2. K. Abraham, in Selected Papers of KarlAbraham (Hogarth, London, 1948), p. 137;S. Freud, in Collected Papers, J. Strachey,Ed. (Hogarth, London, 1956), vol. 4, p. 152.

3. M. Weisman, E. Paykel, G. Klerman, Amer.J. Psychiat. 128, 261 (1971); Y. Cohen, inSocial Structure and Personality, Y. Cohen,Ed. (Holt, Rinehart & Winston, New York,1961), p. 477.

4. R. Spitz. Psychoanal. Study Child 2, 213(1946); ibid. 1, 53 (1945).

5. J. Robertson and J. Bowlby, Cour. Cent.Int. Enfant 2, 131 (1952); J. Bowlby, Int. J.Psychoanal. 41, 89 (1960); ibid. 42, 317 (1961).

6. E. Bibring, in Affective Disorders, P. Green-acre, Ed. (International University Press,New York, 1953), p. 13; W. Gaylin, in TheMeanitng of Despair, W. Gaylin, Ed. (ScienceHouse, New York, 1968), p. 3.

7. L. Ullman and L. Krasner, A PsychologicalApproach to Abnormal Behavior (Prentice-Hall, Englewood Cliffs, N.J., 1969), p. 414;P. Liberman and D. Raskin, Arch. Gen.Psychiat. 24, 515 (1971).

8. A. Prange, Dis. Nerv. Syst. 25, 217 (1964);J. Schildkraut, Amer. J. Psychiat. 122, 509(1965); W. Bunney and J. Davis, Arch. Gen.Psychiat. 13, 483 (1965); A. Coppen, Brit. J.Psychiat. 113, 1237 (1967).

9. P. McGreer, Amer. Sci. 59, 221 (1971).10. G. Klerman, Arch. Gen. Psychiat. 24, 305

(1971).11. P. Whybrow, unpublished manuscript.12. American Psychiatric Association, Diagnostic

and Statistical Manual (American PsychiatricAssociation, Washington, D.C., ed. 2, 1968),p. 35.

28

13. M. Leff, J. Roatch, W. Bunney, Psychiatry33, 293 (1970); E. Paykel, J. Myers, M.Dienelt, G. Klerman, Arch. Gen. Psychiat.21, 753 (1969).

14. D. Graham, Psychosom. Med. 29, 52 (1967);A. Lazare, N. Engl. J. Mcd. 288, 345 (1973).

15. E. Robins and S. Guze, in Recent Advancesin the Psychobiology of Depressive Illness,T. Williams, M. Katz, J. Shields, Eds. (Gov-ernment Printing Office, Washington, D.C.,1972), p. 283.

16. S. Guze, R. Woodruff, P. Clayton, Psychol.Med. 1, 426 (1971); G. Winokur, Dis. Nerv.Syst. 33, 94 (1972).

17. G. Lundquist, Acta Psychiat. Neurol. Scand.Suippi. 35, 1 (1945); C. Perris, Acta Psychiat.Scand. 44, 238 (1968).

18. C. Perris, Acta Psychiat. Scand. Suppl. 42 (No.194), 7 (1966); G. Winokur, P. Clayton, T.Reich, Manic-Depressive Illness (Mosby, St.LoUis, 1969); R. Cadoret, G. Winokur,P. Clayton, Brit. J. Psychiat. 116, 625 (1970);E. Gershon, D. Dunner, F. Goodwin, Arch.Gen. Psychiat. 25, 1 (1971); R. Cadoret andG. Winokur, Int. J. Ment. Health 1, 159 (1972).

19. W. Bunney, D. Murphy, F. Goodwin, Lancet1970-II, 1022 (1970); D. Murphy, H. Brodie,F. Goodwin, W. Bunney, Nature 229, 135(1971).

20. F. Goodwin, D. Murphy, D. Dunner, W.Bunney, Amer. J. Psychiat. 129, 44 (1972).

21. M. Buchsbaum, F. Goodwin, D. Murphy,G. Borge, ibid. 128, 25 (1971); F. Borge, M.Buchsbaum, F. Goodwin, D. Murphy, J.Silverman, Arch. Gen. Psychiat. 24, 501 (1971).

22. A. Beigel and D. Murphy, Arch. Gen.Psychiat. 24, 215 (1971).

23. T. Reich and G. Winokur, J. Nerv. Ment.Dis. 151, 60 (1970).

24. T. Reich, P. Clayton, G. Winokur, Amer. J.Psychiat. 125, 1358 (1969); G. Winokur andV. Tana, Dis. Nerv. Syst. 30, 89 (1969);J. Mendlewicz, J. Fleiss, R. Fieve, J. Amer.Med. Assoc. 222, 1624 (1972); R. Green,V. Goetzl, P. Whybrow, R. Jackson, ibid.223, 1289 (1973).

25. G. Winokur, R. Cadoret, J. Dorzab, M.Baker, Arch. Gen. Psychiat. 24, 135 (1971);G. Winokur, T. Reich, J. Rimmer, F. Pitts,ibid. 23, 104 (1970).

26. M. Blumenthal, ibid. 24, 524 (1971).27. W. McKinney and W. Bunney, ibid. 21, 240

(1969).28. M. Seligman, in The Psychology of Depres-

sion: Contemporary Theory and Research,R. Friedman and M. Katz, Eds., in press.A shorter version appeared in Psychol. Today

-,)7, 43 (1973).29J W. Seay, E. Hansen, H. Harlow, J. Child~ Psychol. Psychiat. 3, 123 (1962); W. Seay

and H. Harlow, J. Nerv. Ment. Dis. 140, 434(1965).

30. H. Harlow and W. McKinney, J. AutismChild. Schiz. 1, 368 (1971).

31. G. Jensen and C. Toleman, J. Comp. Physiol.Psychol. 55, 131 (1962); R. Hinde, Y. Spencer-Booth, M. Bruce, Nature 210, 1021 (1966);G. Mitchell, H. Harlow, G. Moller, Psy-chonomic Sci. 8, 197 (1967).

32. 1. Kaufman and L. Rosenblum, Science 155,1030 (1967); Psychosom. Med. 29, 648 (1967).

33. W. McKinney, S. Suomi, H. Harlow, Amer.J. Psychiat. 127, 1313 (1971); S. Suomi, H.Harlow, C. Domek, J. Abnorm. Psychol. 76,161 (1970).

34. W. McKinney, S. Suomi, H. Harlow, Arch.Gen. Psychiat. 27, 200 (1972).

35. L. Young, S. Suomi, H. Harlow, W. McKin-ney, Amer. J. Psychiat. 130, 400 (1973).

36. F. Goodwin and W. Bunney, Semin. Psy-chiatry 3, 435 (1971).

37. W. McKinney, R. Eising, E. Moran, S. Suomi,H. Harlow, Dis. Nerv. Syst. 32, 735 (1971).

38. S. Spector, A. Sjoerdsma, S. Udenfriend,J. Pharmacol. Exp. Ther. 147, 69 (1965).

39. D. Redmond, J. Maas, A. Kling, H. De-kirmenjian, Psychosoni. Med. 33, 97 (1971).

40. W. McKinney, S. Suomi, I. Mirsky, R. Miller,in preparation.

41. D. Redmiiond, J. Maas, A. Kling, C. Graham,H. Dekirmenjian, Science 174, 428 (1971).

42. B. Koe and A. Weisman, J. Pharmacol. Exp.Ther. 154, 499 (1966).

43. P. Whybrow and J. Mendels, Amer. J.Psychiat. 125, 1491 (1969).

44. G. Breese and P. Traylor, J. Pharmacol. Exp.Ther. 174, 413 (1970); Brit. J. Pharmacol.42, 88 (1971); N. Uretsky and L. Iverson,J. Neutrochein. 17, 269 (1970).

45. L. Stein and C. Wise, Science 171, 1032(1971).

46. L. Stein, J. Psychiat. Res. 8, 354 (1971).47. J. Howard, L. Grant, G. Breese, Pharmacolo-

gist 13, 233 (1971); G. Breese, J. Howard,J. Leahy, Brit. J. Pharmacol. 43, 255 (1971).

48. G. Breese, A. Prange, J. Howard, M. Lipton,W. McKinney, R. Bowman, P. Bushnell,Nature 240, 286 (1972).

49. R. Grinker, The Phenomena of Depression(Hoeber, New York, 1961); I. Melges andJ. Bowlby, Arch. Gen. Psychiat. 20, 690(1969).

50. A. Beck, Depression: Causes and Treatment(Univ. of Pennsylvania Press, Philadelphia,1967), p. 23; K. Minkoff, E. Bergman, A.Beck, R. Beck, Amer. J. Psychiat. 130, 455(1973).

51. M. Seligman and S. Maier, J. Exp. Psychol.74, 1 (1967); J. Overmier and M. Seligman,J. Comp. Physiol. Psychol. 63, 28 (1967);M. Seligman, S. Maier, J. Geer, J. Abnorm.Soc. Psychol. 73, 256 (1968); M. Seligmanand D. Groves, Psychonomic Sci. 19, 191(1970).

52. W. Bonime, in Amlerican Handbook of Psy-chiatry, S. Arieti, Ed. (Basic Books, NewYork, 1966), vol. 3, p. 239; A. Lazare andG. Klerman, Amer. J. Psychiat. 124 (Suppl.),48 (1968).

53. A. Friedman, J. Abnorm. Soc. Psychol. 69,237 (1964); A. Loeb, A. Beck, S. Feshbach,in Proceedings of the 75th Annual Conven-tion of the American Psychological Associa-tion (American Psychological Association,Washington, D.C., 1967), p. 193.

54. C. Ferster, in Research in Behavior Modifi-cation, L. Krasner and L. Ullman, Eds.(Holt, Rinehart & Winston, New York, 1965),p. 6; A. Lazarus, Behav. Res. Ther. 6, 83(1968); J. Wolpe, Amer. 1. Psychother. 25,362 (1971).

55. P. Lewinsohn and M. Shaffer, J. Consult.Clin. Psychol. 37, 87 (1971); P. Lewinsohnand D. Shaw, Psychother. Psychosom. 17, 82(1969).

56. E. Burgess, in Advances in Behavior Therapy,R. Rubin and C. Franks, Eds. (AcademicPress, New York, 1969), p. 53.

57. J. Schildkraut, N. Engl. J. Med. 281, 197,248, 302 (1969).

58. A. Coppen, P. Whybrow, R. Noguera, R.Maggs, A. Prange, Arch. Gen. Psychiat. 26,234 (1972).

59. A. Coppen, D. Shaw, M. Harrell, Lancet1963-H, 527 (1963); C. Pare, ibid., p. 527;A. Coppen, D. Shaw, B. Hertzberg, ibid.1967-II, 1178 (1967); A. Glassman and S.Platman, J. Psychiat. Res. 7, 83 (1969).

60. G. Klerman, J. Schildkraut, J. Hassenbush,J. Psychiat. Res. 1, 289 (1963); F. Goodwin,D. Murphy, H. Brodie, W. Bunney, Biol.Psychiat. 2, 341 (1970).

61. A. Coppen, A. Prange, P. Whybrow, R.Noguera, Arch. Gen. Psychiat. 26, 474 (1972);A. Coopen, J. Psychiat. Res. 9, 1963 (1972).

62. F. Goodwin, Semin. Psychiatry 3, 477 (1971).63. D. Dunner, E. Gershon, Life Sc.

10, 751 (1971); D. Dunner and F. Goodwin,Arch. Gen. Psychiat. 26, 364 (1972).

64. D. Shaw, F. Camps, E. Eccleston, Brit. J.Psychiat. 113, 1407 (1967); H. Bourne, W.Bunney, R. Colburn, J. Davis, J. Davis,Lancet 1968-HI, 805 (1968); C. Pare, D. Yeung,K. Price, R. Stacy, ibid. 1969-H, 133 (1969).

65. G. Ashcroft, T. Crawford, E. Eccleston,Lancet 1966-H, 1049 (1966); S. Dencker, V.Malm, B. Roos, B. Werdinius, J. Neurochem.13, 1545 (1966).

66. H. Van Praag, J. Korf, J. Puite, Nature 225,1259 (1970); F. Goodwin, R. Post, D. Dunner,Amer. J. Psychiat. 130, 73 (1973); H. VanPraag, J. Korf, D. Schut, Arch. Gen. Psychiat.28, 827 (1973).

67. J. Maas and D. Landis, J. Pharmacol. Exp.Ther. 163, 147 (1968); S. Schanberg, J.Schildkraut, I. Kopin, Biochem. Pharmacol.17, 247 (1968); J. Maas and D. Landis,Psychosom. Med. 28, 247 (1966).

68. J. Fawcett, J. Maas, H. Dekirmenjian, Arch.Gen. Psychiat. 26, 246 (1972).

69. M. Ebert, R. Post, F. Goodwin, Lancet1972-H, 766 (1972); R. Post, E. Gordon,F. Goodwin, W. Bunney, Science 179, 1002(1973); R. Post, J. Kotin, F. Goodwin, E.Gordon, Amer. J. Psychiat. 130, 73 (1973).

70. D. Murphy, Amer. J. Psychiat. 129, 141(1972); J. Axelrod, Semin. Psychiatry 4, 199(1972).

71. A. Prange, paper presented at National In-stitute of Mental Health Conference onserotonin and behavior, organized by Stan-

SCIENCE, VOL. 182

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

ford University and held in Palo Alto, Cali-fornia, January 1972; J. Court, Brit. J.Psychiat. 120, 133 (1972).

72. W. Bunney, F. Goodwin, D. Murphy, Arch.Gen. Psychiat. 27, 312 (1972); J. Psychiat.Res. 9, 207 (1972).

73. D. Janowsky, M. El-Yousef, J. Davis,Lancet 1972-H, 632 (1972).

74. A. Prange, I. Wilson, A. Knox, J. Psychiat.Res. 9, 187 (1972).

75. A. Prange, I. Wilson, A. Rabon, M. Lipton,Amer. J. Psychiat. 126, 457 (1969); A. Prange,I. Wilson, A. Knox, T. McClane, M. Lipton,ibid. 127, 191 (1970); D. Wheatley, Arch. Gen.Psychiat. 26, 229 (1972); A. Coppen, P. Why-brow, R. Noguera, R. Maggs, A. Prange,ibid., p. 234.

76. M. Harlow and H. Harlow, Discovery 27, 11(1966).

77. J. Dollard and N. Miller, Personality andPsychotherapy (McGraw-Hill, New York,1950), p. 132.

78. For a comprehensive review of object rela-tions, attachment, and love, see M. Ains-worth, Child Develop. 40, 969 (1969).

79. H. Harlow, Amer. Psychol. 13, 673 (1958);and M. Harlow, Amer. Scf. 54, 244

(1966).80. L. Stein, in Recent Advances in Biological

Psychiatry, J. Wortis, Ed. (Plenum, NewYork, 1962), vol. 4, p. 288; L. Stein, inPsychopharmacology: A Review of Progress,1957-1967, D. Efron, Ed. (Government Print-ing Office, Washington, D.C., 1968), p. 105.

81. J. Olds and P. Milner, J. Comp. Physiol.Psychol. 47, 419 (1954); J. Delgado, W.Roberts, N. Miller, Amer. J. Physiol. 179,587 (1954); R. Heath, Ed. The Role ofPleasure in Behavior (Hoeber, New York,1964).

82. C. Wise, B. Berger, L. Stein, Biol. Psychiat.6, 3 (1973). Stein and his associates havepostulated that the biochemical etiology of allthe major psychiatric disorders is to be foundin derangements in MFB and PVS (45,46, 80).

83. M. Jouvet, Science 163, 32 (1969).84. B. Poschell, Physiol. Behav. 3, 53 (1967).85. A. Coppen and D. Shaw, Brit. Med. J. 2,

1439 (1963).

86. D. Bogdanski and B. Brodie, Life Sc. 5,

1563 (1966).87. For a discussion of mineral metabolism,

catecholamines, and depression, see J. Maas,J. Psychiat. Res. 9, 227 (1972).

88. J. Weiss, J. Comp. Physiol. Psychol. 65, 251(1968); , E. Stone, N. Harrel, ibid. 72,

153 (1970).89. K. L. Granville-Grossman, in Recent De-

velopments in Affective Disorders, A. Coppenand A. Walk, Eds. (Headley, Ashford, Eng-land, 1968), p. 65.

90. D. Robinson, J. Davis, A. Nies, C. Ravaris,D. Sylvester, Arch. Gen. Psychiat. 24, 536(1971).

91. We thank L. Benjamin, P. Lang, and J.Westman (University of Wisconsin) and E.Brown, G. Aivazian, J. Beard, and B. Kulig(University of Tennessee) for their criticalcomments on earlier versions of the manu-script. This work was supported, in part, byresearch grants MH-18070 and MH-21892and research scientist development awardMH-47353 (W.T.M.), all from the NationalInstitute of Mental Health.

Computer and Information Networks

The movement in research and education toward

national resource sharing via networks is accelerating.

Martin Greenberger, Julius Aronofsky,James L. McKenney, William F. Massy

A medical researcher sits at an on-

line terminal in Honolulu searching an

index to the world's medical literaturestored on a computer in Bethesda,Maryland, over 5000 miles away. Hisrequest passes across a radio networkof the University of Hawaii; the tele-phone network of ithe Hawaiian tele-phone company; the Pacific Ocean viaan international satellite; a nationwideresearch network in the continentalUnited States; and, finally, a commer-

cial time-sharing network, before reach-ing the medical information system inBethesda. By mail the request wouldhave taken several days. By computer-communication networks it takes lessthan 5 seconds. The response to therequest, a set of literature citations,starts printing out at the terminal backin Honolulu within 15 seconds fromthe time the request was dispatched.Numerous other remote users of themedical information system receiveservice simultaneously.

This example of what is happeningnow may seem dramatized, but it

5 OCTOBER 1973

does illustrate the daily use of a varietyof communication networks that are

currently providing efficient intercon-nection between computer systems fortheir users. Domestic usage of the on-

line medical information systemthrough the commercial time-sharingnetwork has been doubling every 6months. This fact plus countless otherimportant uses of networks in many

areas proclaim the growing significanceof information and computer networks.The possibilities they offer in researchand education point to new and bettercomputing and information services,greater efficiency in operations, broadermarkets, widespread access to facilities,and extensive resource sharing.

Responding to the heightened inter-est in the possibilities of networks, andreflecting its own continuing interest inimproving the use of new technologiesin research and education- (1), the Na-tional Science Foundation (NSF) in1972 announced the mounting of "anexpanded research program . . . toexplore . . . the resource-sharing po-

tential of a national network in sup-port of research and education" (2).An NSF grant under this program per-mitted EDUCOM to bring togetherinterested users and administratorswith those possessing shareable re-sources and relevant experience in aseries of three 2-day working seminars.The seminars, held in late 1972 andearly 1973, were designed to helpidentify the central organizational,political, and economic issues in build-ing and operating networks on a na-tional basis.EDUCOM has been concerned

since its founding in 1964 with foster-ing the collaboration of colleges anduniversities in the use of computer andcommunication technologies. It hasgiven the subject of networks specialemphasis, beginning with its July 1966summer study in Boulder, Colorado(3), and continuing with the openconferences that it recently has beenholding twice each year (4). In theseworking seminars, highly expert tech-nologists joined in discussion withsocial scientists, physical scientists,decision-makers, and others from manyfields (5). This article is based on theresults of their deliberations (6).

This article is adapted from the book Networksfor Research and Education-Sharing Computerand Information Resources Nationwide (MITPress, Cambridge, Mass., in press). It reportson a series of working seminars conducted bythe authors for EDUCOM with National ScienceFoundation support. Dr. Greenberger, directorof the seminars, is professor of mathematicalsciences and senior staff associate of the Centerfor Metropolitan Planning and Research at JohnsHopkins University, Baltimore, Maryland. Dr.Aronofsky is professor of management scienceand computing at Southern Methodist University,Dallas, Texas. Dr. McKenney is professor ofbusiness administration at Harvard University,Cambridge, Massachusetts. Dr. Massy is vice pro-vost for research and professor of business ad-ministration at Stanford University, Stanford,California.

29

on

Apr

il 4,

200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from