Affectional Response inthe InfantMonkey

21 August 1959, Volume 130, Number 3373

SCIENCE

CURRENT PROBLEMS IN RESEARCH

Orphanedbabymonkeysdevelopa strongandpersisterit

attachment to inanimate surrogate mother S.

In Harry F. Harlow and Robert R. Zimmerman

Investigatorsfromdiversebehavioral thatmonkeyandchimpanmzeeinfantsde- fields have long recognized the strong velop strong ties to their mothers and attachmentoftheneonatalandinfantile thattheseaffectionalattaichmentsmay animaltoitsmother.Althoughthisaf- persistforyears.Itis,ofcourse,com- fectionalbehaviorhasbeencommonly monknowledgethathiumaninfants observed,thereis,outsidethefieldof formstrongandpersistenttiestotheir ethology,scantexperimentalevidence mothers. permittingidentificationofthefactors Althoughstudentsfromdiversescien- criticaltotheformationofthisbond. tificfieldsrecognizethisa.bidingattach- Lorenz(1)andothershavestressedthe ment,thereisconsiderabledisagreement importanceofinnatevisualandauditory aboutthenatureofitsdevrelopmentand mechanismswhich,throughtheprocess itsfundamentalunderlyingmechanisms. ofimprinting,giverisetopersistingfol- -A-commontheoryamongpsychologists, lowingresponsesintheinfantbirdand sociologists,andanthropologistsisthat fish.Imprintingbehaviorhasbeendem- oflearningbasedondrivereduction. onstratedsucesfulyinavariety.of Thistheoryproposesthattheinfant’s avianspeciesundercontrolledlabora- attachmenttothemotherrresultsfrom toryconditions,andthisphenomenon theassociationofthemotlher’sfaceand hasbeeninvestigatedsystematicallyin formwiththealleviation(ofcertainpri- order to identify those variables which mary drive states, particiularly hunger contributetoitsdevelopmentandmain- ,andthirst.Thus,throughlearning,af-

preconceived theoretical framework. An exception to the above generalization is seen in the recent attempt by Bowlby (14) to analyze and integrate the avail- able observational and experimental evi- dence derived from both human and subhumaninfants.Bowlbyhasconcluded thatatheoryofcomponentinstinctual responses, species specific, can best ac- count for the infant’s tie to the mother.

He suggeststhatthespecies-specificre- sponses for human beings (some of these

responses are not strictly limited to hu- man beings) include contact, clinging, sucking, crying, smiling, and following. He further emphasizes that these re- sponses are manifested independently of primary drive reduction in human and subhuman infants.

Theabsenceofexperimentaldata which would allow a critical evaluation ofanytheoryofaffectionaldevelopment canbeattributedtoseveralcauses.The useofhumaninfantsassubjectshasseri- ouslimitations,sinceitisnotfeasibleto employ al the experimental controls whichwouldpermitacompletelyade- quateanalysisoftheproposedvariables. Inaddition,thelimitedresponsereper- toireofthehumanneonateseverelyre- strictsthenumberofdiscreteorprecise responsecategoriesthatcanbemeas- ureduntilaconsiderableagehasbeen attained.Thus,criticalvariablesgoun-

measuredandbecomelostorconfounded amongthecomplexphysiological,psy- chological,andculturalfactorswhich influencethedevelopinghumaninfant.

Moreover, the use of common labora- toryanimalsalsohasseriouslimitations, formostoftheseanimalshavebehav- ioral repertoires very different from thoseofthehumanbeing,andinmany speciesthesesystemsmaturesorapidly thatitisdificulttomeasureandassess theirorderlydevelopment.On theother hand, subhuman primates, including the macaque monkey, are born at a state of maturity which makes it possible to

ObservationsonmonkeysbyCarpen-
ter(7),Nolte(8),andZuckermann tionalresponseshasledthesetheorists firstfewdaysoflife.Furthermore,their

tenance[se,forexample,Hinde, Thorpe, and Vince (2), Fabricius (3), Hess(4),Jaynes(5),andMoltzand Rosenblum(6)].Thesestudiesrepresent thelargestbodyofexistentexperimen- tal evidence measuring the tie between infant and mother. At the mammalian levelthereislitleorno systematicex- perimental evidence of this nature.

fectionbecomesaself-supporting,de- rived drive (12). Psychoarialysts, on the otherhand,havestressedtiheimportance ofvariousinnateneeds,suichasaneed tosuckandorallypossesthebreast (2), or needs relating to c,ontact, move- ment, temperature (13), and clinging to the mother (14).

(9)andonchimpanzeesbyKohler(10) and by Yerkes and Tomilin (11) show

The authors are on the staff of the primate laboratory, department of psychology, University ofWisconsin,Madison.

21AUGUST1959

toderivetheirbasic ?othesesfront hyp

deductions and intuitions based on ob- servationandanalysisofadultverbal reports. As a result, the asiailable obser- vational evidence is often forced into a

postnatalmaturationalrateisslow enough to permit precise assessment of affectionalvariablesanddevelopment.

Over a 3-year period prior to the be- ginning of the research program reported

The paucity of experime ental evidence
concerningthedevelopmentofaf&e beginprecisemeasurementswithinthe

, 421

here (15), some 60 infant macaque monkeyswereseparatedfromtheir mothers 6 to 12 hours after birth and raised at the primate laboratory of the UniversityofWisconsin.Thesuccessof the procedures developed to care for theseneonateswasdemonstratedbythe low mortality and by a gain in weight which was approximately 25 percent greaterthanthatofinfantsraisedby their own mothers. All credit for the success of this program belongs to van Wagenen (16), who had described the essential procedures in detail.

capabilities of the neonatal and infantile
monkey.Thestudieswhichresultedhave velopmentoftheinfant’sattachmentto

24 21 18

U)15 v12

Z 9 6 3 O

422

18 ‘5

Fig. 1. Wire and cloth mother surrogates.

with, and are considered basic to, affec- tion; these include nursing, clinging, and visual and auditory exploration.

In the course of raising these infants we observed that they al showed a strong attachment to the cheesecloth blankets which were used to cover the wire floors of their cages. Removal of these cloth blankets resulted in violent emotional behavior. These responses were not short-lived; indeed, the emo- tional disturbance lasted several days, as was indicated by the infant’s refusal to work on the standard learning tests that were being conducted at the time. Simi- lar observations had already been made by Foley (17) and by van Wagenen (16), who stressed the importance of adequate contact responses to the very survival ofthe neonatal macaque. Such observations suggested to us that con- tact was a true affectional variable and that it should be possible to trace and measure the development and impor- tance of theseresponses. Indeed there seemed to be every reason to believe that one couldmanipulatealvariableswhich have been considered critical to the de-

These first 3 years were spent in de-
visingmeasurestoassessthemultiple isticresponsesthathavebeenassociated

1-5 6- 1I-s 16-20 21-25 MEAN AGE

O—WIRE MOTHER a_=-=:__-+-

5 25 85 105 125 145 165 MEAN AGE

revealed that the development of per- ception, learning, manipulation, explo- ration,frustration,andtimidityinthe macaque monkey follows a course and sequencewhichisverysimilartothatin the human infant. The basic differences between the two species appear to be theadvancedpostnatalmaturationalsta- tus and the subsequent more rapid growth of the infant macaque. Probably the most important similarities between the two, in relation to the problem of affectional development, are character-

a mother, or mother surrogate.
To attain control over maternal vari- ables,wetookthecalculatedriskofcon- structing and using inanimate mother

surrogatesratherthanrealmothers.The cloth mother that we used was a cylin- der of wood covered with a sheath of terrycloth(18),andthewiremother was a hardware-cloth cylinder. Initialy, sponge rubber was placed underneath the terry cloth sheath of the cloth mother surrogate, and a light bulb be- hind each mother surrogate provided radiantheat.Forreasonsofsanitation

~CLOTHFED_
A ~~CLOTHMOTHER

UI)

121 _ // //

er _g/ UN~~WIREMOTHER _ 9

0Ir

6 31

~WIREFE
o-o CLOTH MOTHER

Time spent on cloth and wire mother surrogates. Fig. 2 (left). Short term. Fig. 3 (right). Long term.
SCIENCE, VOL. 130

and safety these two factors were elimi- nated in construction of the standard mothers, with no observable effect on the behavior of the infants. The two mothers were attached at a 45-degree angle to aluminum bases and were given different faces to assure uniqueness in the various test situations (Fig. 1). Bot- tle holders were installed in the upper middle part of the bodies to permit nurs- ing. The mother was designed on the basis of previous experience with infant

affectional attachment. It could merely reflect the fact that the cloth mother is a more comfortable sleeping platform or amore adequate source of warmth for the infant. However, three of the four infants nursed by the cloth mother and one of the four nursed by the wire mother left a gauze-covered heating pad that was on the floor of their cages dur- ing the first 14 days of life to spend up to 18 hours a day on the cloth mother. This suggests that differential heating or warmth is not a critical variable within the controlled temperature range of the laboratory.

monkeys, which suggested that nursing
in an upright or inclined position with
something for the infant to clasp facili-
tated successful nursing and resulted in
healthier infants (see 16). Thus, both
mothers provided the basic known re-
quirements for adequate nursing, but
the cloth mother provided an additional
variable of contact comfort. That both
of these surrogate mothers provided ade-
quate nursing support isshown by the
factthatthetotalingestionofformula erence fortheclothmother.Theper- sponsivenessofthemonkeyinfantsunder andtheweightgainwasnormalforal sistenceofthedifferentialresponsive- conditionsofdistressorfear.

infantsfedonthesurrogatemothers.The ness tothemothersforbothgroups of onlyconsistentdifferencebetweenthe infantsisevident,andtheover-alldif-

Various fear-producing stimuli, such asthemovingtoybearillustratedinFig. 4, were presented to the infants in their home cages. The data on differential responses under both feeding conditions are given in Fig. 5. It is apparent that the cloth mother was highly preferred to the wire mother, and it is a fact that thesedifferenceswereunrelatedtofeed- ing conditions-that is, nursing on the cloth or on the wire mother. Above and beyond these objective data are obser-

groupslayinthesofterstoolsofthein- fantsfedonthewiremother.

DevelopmentofAffectionalResponses

ferencesbetweenthegroups fallshortof statistical significance.

These data make itobvious that con- tactcomfortisa variableofcriticalim- portance in the development of affec- tional responsiveness to the surrogate

Theinitialexperimentsonthedevel- mother,andthatnursingappearstoplay opment of affectional responses have a negligible role. With increasing age already been reported (19) but will be and opportunity to learn, an infant fed

Fig. 4. Typical fear stimulus.

Other tests demonstrate that the cloth mother ismore than a convenient nest; indeed, they show that a bond develops between infant and cloth-mother surro- gate that is almost unbelievably similar to the bond established between human mother and child. One highly definitive test measured the selective maternal re-

brieflyreviewedhere,sincesubsequent froma lactatingwiremother-doesnot
experimentswerederivedfromthem.In becomemoreresponsivetoher,aswould vationsontheformoftheinfants’re-

the initial experiments, designed to
evaluate the role of nursing on the de-
velopmentofaffection,aclothmother
and a wire mother were placed in dif-
ferentcubiclesattachedtotheinfant’s variancewitha drive-reductiontheory livingcage.Eightnewbornmonkeys ofaffectionaldevelopment. wereplacedinindividualcageswiththe Theamount oftimespenton the

surrogates; for four infant monkeys the cloth mother lactated and the wire motherdidnot,andfortheotherfour thisconditionwasreversed.

mother does not necessarily indicate an

HOME CAGE FEAR FIRST RESPONSE DUAL FED RAISED

The infants lived with their mother
surrogates for a minimum of 165 days,
and during this time they were tested
in a variety of situations designed to
measure the development of affectional
responsiveness. Differential affectional
responsivenesswasinitialymeasuredin
terms of mean hours per day spent on
the cloth and on the wire mothers under
two conditions of feeding, as shown in
Fig. 2. Infants fed on the cloth mother
and on the wire mother have highly
similar scores after a short adaptation
period(Fig.3),andovera165-day Fig.5.Differentialresponsivenessinfear periodbothgroupsshowadistinctpref- tests.

21 AUGUST 1959

423

be predicted from a derived-drive the- ory, but instead becomes increasingly more responsivetoitsnonlactatingcloth mother. These findings are at complete

sponsesinthissituation.Inspiteoftheir abject terror, the infant monkeys, after reaching the cloth mother and rubbing their bodies about hers, rapidly come to losetheirfearofthefrighteningstimuli. Indeed, within a minute or two most of thebabieswerevisuallyexploringthe very thing which so shortly before had seemedanobjectofevil.Thebravestof the babies would actually leave the motherandapproachthefearfulmon- sters, under, of course, the protective gaze of their mothers.

These data are highly similar, in terms of differential responsiveness, to thetimescorespreviouslymentionedand indicate the overwhelming importance ofcontactcomfort.The resultsareso striking as to suggest that the primary functionofnursingmaybethat ofin- suring frequent and intimate contact between mother and infant, thus facili- tating the localization of the source of contact comfort. This interpretation finds some support in the test discussed above. In both situations the infants

I100 to
z

CLs8o (0)

LW 60

X. 40

I-
z wo

WL 0

MEAN DAY OF AGE

nursedbytheclothmotherdeveloped motherorturntogazeatthefeared consistent responsiveness to the soft objectwithouttheslightestsignofap- motherearlierintestingthandidthe prehension.Theinfantsraisedonthe infantsnursedbythewiremother,and wiremother,ontheotherhand,rushed duringthistransientperiodthelatter awayfromthefearedobjecttoward groupwasslightlymoreresponsivetothe theirmotherbutdidnotclingtoorem- wiremotherthantheformergroup. braceher.Instead,theywouldeither However,theseearlydifferencesshortly

days (see Fig. 5). These data suggest the possible hypothesis that nursing fa- cilitated the contact of infant and mother during the early developmental periods.

Theinterpretationofalfeartesting iscomplicatedbythefactthatalor most”fear”stimulievokemanypositive exploratory responses early in life and donotconsistentlyevokeflightresponses untilthemonkeyis40to50daysofage. Similardelayedmaturationofvisually inducedfearresponseshasbeenreported

clutchthemselvesandrockandvocalize
for the remainder of the test or rub
againstthesideofthecubicle.Contact
withthecubicleorthemotherdidnot
reducethe&notionalityproducedbythe
introductionofthefearstimulus.These
ing-cage.Fouroftheinfantswerepre- differencesarerevealedinemotionality forbirds(3),chimpanzees(20),and

disappeared. Additionaldatahavebeenobtained

fromtwogroupsoffourmonkeyseach whichwereraisedwithasinglemother placedinacubicleattachedtotheliv-

sentedwithalactatingwiremotherand scores,forbehaviorsuchasvocalization, humaninfants(21).
the other four were presented with a crouching, rocking, and sucking, re- Because of apparent interactions be- nonlactatingclothmother.Thelatter cordedduringthetest.Figure8shows tweenfearfulandaffectionaldevelop- groupwashand-fedfromsmallnursing themeanemotionalityindexfortest mentalvariables,atestwasdesignedto bottlesforthefirst30daysoflifeand sessionsforthetwoexperimentalgroups, tracethedevelopmentofapproachand then weaned to a cup. The develop- the dual-mother groups, and a compar- avoidance responses in these infants. mentofresponsivenesstothemothers ablecontrolgroupraisedunderstandard Thistest,describedasthestraight-alley

wasstudiedfor165days;afterthisthe laboratoryconditions.Ascanbeseen,

test,wasconductedinawoodenalley8 feet long and 2 feet wide. One end of the alley contained a movable tray uponwhichappropriatestimuliwere placed. The other end of the alley con- tainedaboxforhiding.Eachtestbegan withthemonkeyinastartbox1footin frontofthehidingbox;thus,theanimal

individual mothers were removed from the cages and testing was continued to determinethestrengthandpersistence of the affectional responses.

the infants raised with the single wire mother have the highest emotionality scoresofalthegroups,andtheinfants raised with the single cloth mother or withaclothandwiremotherhavethe lowestscores.Itappearsthatthere- sponsesmadebyinfantsraisedonlywith

Figure6presentsthemeantimeper
dayspentontherespectivemothersover
the165-daytestperiod,andFig.7shows
thepercentageofresponsestothemoth- awiremotherweremoreinthenature couldmaintainhisoriginalposition,ap-

erswhenafear-producingstimuluswas ofsimpleflightresponsestothefear introducedintothehomecage.These stimulusandthatthepresenceofthe testsindicatethatbothgroupsofinfants mothersurrogatehadlitleeffectin developedresponsivenesstotheirmother alleviatingthefear.

proachthestimulustrayasitmoved towardhim,orfleintothehidingbox. Theinfantswerepresentedwithfive stimuli in the course of five successive days.Thestimuliincludedastandard clothmother,astandardwiremother, ayellowclothmotherwiththeheadre- moved,ablanktray,andalargeblack fear stimulus, shown in Fig. 9. The in- fantsweretestedat5,10,and20days ofage,respectively,andthenat20-day intervalsupto160days.Figure10shows

themeannumberof15-secondtime

surrogates.However,thesemeasuresdid
notrevealthedifferencesinbehavior
thatweredisplayedinthereactionsto
themotherswhenthefearstimuliwere feartestsdecreasedwithageandoppor-

_presented. The infants raised on the tunity to learn, while responsiveness to clothmotherwouldrushtothemother thenonlactatingclothmotherincreased. andclingtightlytoher.Followingthis However,therewassomeindicationof initialresponsetheseinfantswouldre- aslightincreaseinfrequencyofresponse laxandeitherbegintomanipulatethe tothewiremotherforthefirst30to60

21 18

U) cr 1 5

0X12

(/100 A, al W

O 80

0~ \ n-

W .0′ 6

SINGLECLOTHNONFED – O-0 SINGLE WIRE FED

—

3 0

l1 LI A A A A A A 20 45 65 85 lOS 125 145 165 MEANDAYSOFAGE

TIME SPENT ON CLOTH AND WIRE MOTHERS

HOME CAGE FEAR WITHDRAWAL TO MOTHER

p-~~~~~~~~~

,M _~~~~~~~~~~

c20 -@-*–*SINGLECLOTHNONFED –

a-0bo ~~CO–SINGLEWIREFED

_, ,,, 1 A

5-10 20-40 60-80 100120 140160 DAYS OF AGE

1 . -1 9.

0- 1

/’

W 60 It z40

Duringourinitialexperimentswith thedual-motherconditions,responsive- nesstothelactatingwiremotherinthe

0-~~~~~~’D —0

Fig.6(left).Timespenton singlemothersurrogates.Fig.7(right).Responsivenesstosinglesurrogatemothersinfeartests.
424 SCIENCE, VOL. 130

HOME CAGE FEAR
TOTAL EMOTION SCORE INCREASE FROM NORM-MECHANICAL STIMULI

cate that the cloth mother provides a haven of safety and security for the frightened infants The affectional re- sponsepatternsfoundintheinfantmon- key are unlike tropistic or even complex reflex responses; they resemble instead the diverse and pervasive patterns of re- sponse to his mother exhibited by the human child in the complexity of situa- tions involving child-mother relation- ships.

The role of the mother as a source of safety and security has been demon- stratedexperimentallyforhuman infants by Arsenian (22). She placed children 11to30monthsofageinastrangeroom containing toys and other play objects. Half of the children were accompanied into the room by a mother or a substi-

DUAL SINGLE SINGLE CONTROLS FED CLOTH WIRE

NON-FED FED

Fig. 8. Change in emotionality index in feartests.

Fig. 9. Response to the fear stimulus in thestraight-alleytest.

fants were given repetitive tests to ob- tain information on the development of responsivenesstoeachofthedualmoth- ersinthissituation.A clothblanketwas always available as one of the stimuli throughoutthesesions.Itshouldbeem- phasized that the blanket could readily compete with the cloth mother as a con- tact stimulus, for it was standard labora- tory procedure to wrap the infants in soft cloth whenever they were removed from their cages for testing, weighing, and other required laboratory activities.

As soon as they were placed in the test room, the infants raised with cloth mothers rushed to their mother surro- gate when she was present and clutched hertenaciously,aresponsesostrongthat it can only be adequately depicted by motion pictures. Then, as had been ob- servedinthefeartestsinthehome cage, they rapidly relaxed, showed no sign of apprehension, and began to demonstrate unequivocal positive responses of ma- nipulating and climbing on the mother. After several sessions, the infants began to use the mother surrogate as a base of

TIME IN CONTACT WITH MOTHER

ent during one of the weekly sessions
and the appropriate mother surrogate
(the kind which the experimental infant
Thedifferentialresponsivenesstothe hadalwaysknown)beingpresentdur-
cloth mother of infants raised with ingtheothersesions.Fourinfantsraised
bothmothers,thereducedemotionality withdualmothersurrogatesandfour
ofboththegroupsraisedwithcloth controlinfantsweresubjectedtosimilar
mothersinthehome-cagefeartests, experimentalsequences,theclothmother
and the development of approach re- being present on half of the occasions. Fig. 10. Responsiveness to mother surro-

sponsesinthestraight-aleytestindi- Theremainingfour”duial-mother”in- gatesinthe6traight-alleytests. 21AUGUST 1959

periodsspentincontactwiththeappro- tutemother(afamiliarnurseryattend- priatemotherduringthe90-secondtests ant),whiletheotherhalfenteredthe forthetwosingle-mothergroups,and situationalone.Thechildreninthefirst theresponsestotheclothmotherbyfour group(motherpresent)weremuchles infants from the dual-mother group. emotional and participated much more

Duringthefirst80daysoftesting,al fullyintheplayactivitythanthose thegroupsshowedanincreaseinre- inthesecondgroup(motherabsent). sponse to the respective mother surro- Withrepeatedtesting,thesecurityscore, gates.Theinfantsfedonthesinglewire acompositescoreofemotionalityand mother,however,reachedpeakrespon- playbehavior,improvedforthechil- siveness at this age and then showed a drenwho enteredalone,butitstilfel consistentdecline,followedbyanactual farbelowthatforthechildrenwhowere avoidanceofthewiremother.During accompaniedbytheirmothers.Insub- test sessions 140 to 160, only one contact sequent tests, the children from the wasmadewiththewiremother,and mother-presentgroupwereplacedinthe threeofthefourinfantsranintothe testroomalone,andtherewasadrastic hiding box almost immediately and re- dropinthesecurityscores.Contrariwise, mainedtherefortheentiretestsesion. theintroductionofthemotherraised Ontheotherhand,aloftheinfants thesecurityscoresofchildreninthe

raised with a cloth mother, whether or
not they were nursed by her, showed a
progressive increase in time spent in con-
tactwiththeirclothmothersuntilap- keysraisedonmothersurrogateswith proaches and contacts during the test control monkeys raised in a wire cage sessionsapproachedmaximum scores. containing a cheesecloth blanket from

.Thedevelopmentoftheresponseof days1to14andnoclothblanketsub- flightfromthewiremotherbythegroup sequently.Theinfantswereintroduced fedonthesinglewiremotheris,of intothestrangeenvironmentoftheopen course,completelycontrarytoaderived- field,whichwasaroommeasuring6by drivetheoryofaffectionaldevelopment. 6by6fet,containingmultiplestimuli Acomparisonofthisgroupwiththe knowntoelicitcuriosity-manipulatory groupraisedwithaclothmothergives responsesinbabymonkeys.Theinfants somesupporttothehypothesisthatfeed- raisedwithsinglemothersurrogateswere ingornursingfacilitatestheearlydevel- placedinthissituationtwiceaweekfor opmentofresponsestothemotherbut 8weeks,nomothersurrogatebeingpres-

that without the factor of contact com- fort, these positive responses are not maintained.

other group.
We haveperformedasimilarseriesof

open-fieldexperiments,comparingmon-

DAYS OF AGE

425

Fig.1.Subsequentresponsetoclothmotherandstimulusintheopen-fieldtest.

glewiremotherswas bothquantitatively and qualitatively different from that of the infants raised with cloth mothers. Not only did these infants spend litle orno timecontactingtheirmothersur- rogates but the presence of the mother did not reduce their emotionality. These differencesareevidentinthemean num- beroftimeperiodsspentincontactwith the respective mothers, as shown in Fig. 13, and the composite emotional index forthetwo stimulusconditionsdepicted in Fig. 14. Although the infants raised with dual mothers spent considerably more time in contact with the cloth mother than did the infants raised with single cloth mothers, their emotional re- actions to the presence and absence of the mother were highly similar, the com- posite emotional index being reduced by almost half when the mother was in the test situation. The infants raised with wire mothers were highly emotional under both conditions and actually showed a slight, though nonsignificant, increaseinemotionalitywhenthemother was present. Although some of the in- fantsrearedbyawiremotherdidcon- tacther,theirbehaviorwassimilarto thatobservedinthehome-cagefear

operations,leavinghertoexploreand crouchedposition,asillustratedinFig.
handleastimulusandthenreturningto 12,orrunningaroyndtheroomonthe
herbeforegoingtoanewplaything. hindfet,clutchingthemselveswith
Someoftheinfantsevenbroughtthe theirarms.Thoughnoquantitativeevi- tests.Theydidnotclutchandclingto stimulitothemother,asshowninFig. denceisavailable,contactandmanipu-

1.Thebehavioroftheseinfants lationofobjectswasfranticandof changedradicallyintheabsenceofthe shortduration,asopposedtotheplayful mother.Emotionalindicessuchasvo- typeofmanipulationobservedwhenthe calization,crouching,rocking,andsuck- motherwaspresent. ingincreasedsharply.Typicalresponse Inthepresenceofthemother,the

theirmotherasdidtheinfantswith clothmothers;instead,theysatonher lapandclutchedthemselves,orheld theirheadsandbodiesintheirarmsand engaged in convulsive jerking and rock- ingmovementssimilartotheautisticbe- havior of deprived and institutionalized human children. The lack of exploratory and manipulatory behavior on the part of the infants reared with wire mothers, both in the presence and absence of the wiremother,wassimilartothatobserved in the mother-absent condition for the infants raised with the cloth mothers, and such contact with objects as was made was of short duration and of an erratic and frantic nature. None of the infants raised with single wire mothers displayed the persistent and aggressive play behavior that was typical of many of the infants that were raised with cloth mothers.

The fourcontrolinfants,raisedwith- out a mother surrogate, had approxi- mately the same emotionality scores when the mother was absent that the other infants had in the same condition, but the control subjects’ emotionality scores were significantly higher in the presence of the mother surrogate than in her absence. This result is not surpris- ing, since recent evidence indicates that the cloth mother with the highly orna- mental face is an effective fear stimulus

SCIENCE, VOL. 130

patterns were

either

freezing

in a

behavior of the infants raised with sin-

Fig.12.Responseintheopen-fieldtestintheabsenceofthemothersurrogate. 42.6

)9 0

w5 0 0

MEAN TIME PERIODS IN CONTACT WITH MOTHER

OPEN FIELD COMPOSITE EMOTIONAL INDEX

z E4K 4

08
W7 -4

W,0~6 0

w3 o I- cca 1-2

:5X Wa 3.- Xz(

Z4 _i 0 W ~

U X
W3 z0k_ z1-0914 J0

SINGLE CLOTH WIRE NONFED FED

— -^ CONTROL

DUAL FED

SINGLE NONFED FED

CONTROL

DUAL FED

SINGLE

x 0~ ~ ~~:x_
04i WX 4

2 U cc 0-

Lu J

Fi _4 U-

1A ra.-

SINGLE
CLOTH WIRE

Fig. 13 (left). Responsiveness to mother surrogates in the open-field test. Fig. 14 (right). Emotionality index in testing with and with- outthemothersurrogates.

formonkeysthathavenotbeenraised forthethreestimulusconditionsduring showingasharpincrease.Theheight- thesesametests.Thedifferentialrespon- enedemotionalityfoundunderthewire- Furtherillustrationofdifferentialre- sivenesstotheclothandwiremothers, motherconditionwasmainlycontrib- utedbythetwoinfantsfedonthewire mother.Thebehaviorofthesetwoin- fantsinthepresenceofthewiremother wassimilartothebehavioroftheani- malsraisedwithasinglewiremother. Onthefewoccasionswhencontactwith the wire mother was made, the infants didnotattempttoclingtoher;instead scoresfromthe45thdayonward.How- theywouldsitonherlap,clasptheir headsandbodies,androckbackand

sponsivenesstothetwomothersurro- gatesisfoundintheresultsofaseries ofdevelopmentaltestsintheopen-field situation,giventotheremainingfour “dual-mother”infants.Theseinfants wereplacedinthetestroomwiththe cloth mother, the wire mother, and no motherpresentonsuccessiveoccasions atvariousagelevels.Figure15shows themeannumberoftimeperiodsspert incontactwiththerespectivemothers fortwotrialsateachagelevel,andFig. 16 reveals the composite emotion scores

asmeasuredbycontacttime,isevident by20daysofage,andthissystematic differencecontinuesthroughout140days ofage.Onlysmalldifferencesinemo- tionalityunderthevariousconditions areevidentduringthefirst85daysof age, although the presence of the cloth motherdoesresultinslightlylower

z2 WU

/~~~~~~~

z 41

MEAN TIME PERIODS IN CONTACT WITH MOTHER

DUAL FED RAISED AII1

COMPOSITE EMOTIONAL INDEX DUAL FED RAISED

U,) 7 – 0-0 WIRE MOTHER u)3 0

0
6

5
CL W-

*-@ CLOTH MOTHER

I–

A/
0 ‘o-

A l a A Aa a 10 25 45 65 85 105

DAYS OF AGE DAYS OFAGE

145

a.laaa.a a 25 45 65 85 105 145

PI -,

ever,at105and145daysofagethere isaconsiderabledifferenceforthethree conditions,theemotionalityscoresfor the wire-mother and blank conditions

forth. In1953Butler(23)demonstrated

that mature monkeys enclosed in a

l~o~~~2~5~104MEAN

DAY

OF AGE

Fig.15 (left).Differentialresponsivenessintheopen-fieldtest.Fig.16(right).Emotionalityindexunderthreeconditionsintheopen- fieldtest.

21 AUGUST 1959 427

IIII` – Ag1

– CLOTHMOTHER *— WIRE MOTHER 0—- BLANK

dimlylightedboxwouldopenandre- dicatingthatfeedingprobablyfacilitated open a door for hours on end with no the early appearance and increased the other motivation than that oflooking early strength of affectional responsive- outsidethebox.Healsodemonstrated ness.Certainlyfeeding,incontrastto thatrhesusmonkeysshowedselectivity contactcomfort,isneitheranecessary inrateandfrequencyofdoor-opening norasufficientconditionforaffectional inresponsetostimuliofdifferentde- development.

RetentionofAffectionalResponses

One oftheoutstandingcharacteristics oftheinfant’sattachmenttoitsmother isthepersistenceoftherelationshipover aperiodofyears,eventhoughthefre- quency of contact between infant and mother is reduced with increasing age. In order to test the persistence of the responsivenessofour”mother-surrogate” infants, the first four infant monkeys raised with dual mothers and al of the monkeys raised with single mothers were separated from their surrogates at 165 to 170 days of age. They were tested for affectional retention during the follow- ing 9 days, then at 30-day intervals dur- ing the following year. The results are of necessity incomplete, inasmuch as theentiremother-surrogateprogramwas initiated les than 2 years ago, but enough evidence is available to indicate thattheattachmentformedtothecloth mother during the first 6 months of life is enduring and not easily forgotten.

Affectional retention as measured by the modified Butler box for the first 15 months of testing for four of the infants raised with two mothers is given in Fig. 18.Althoughthereisconsiderablevaria- bility in the total response frequency from session to session, there is a con- sistent difference in the number of re- sponsestotheclothmotherascontrasted withresponsestoeitherthewiremother or the empty box, and there is no con- sistent difference between responses to the wire mother and to the empty box. Theefectsofcontactcomfortversus feeding are dramatically demonstrated in this test by the monkeys raised with either single cloth or wire mothers. Fig- ure 19 shows the frequency of response to the appropriate mother surrogate and to the blank box during’the presepara- tion period and the first 90 days of re- tention testing. Removal of the mother resulted in a doubling of the frequency of response to the cloth mother and more than tripled the difference be- tween the responses to the cloth mother and those to the empty box for the in- fants that had lived with a single non- lactating cloth mother surrogate. The infantsraisedwithasinglelactatingwire mother, on the other hand, not only failed to show any consistent preference for the wire mother but also showed a highly significant reduction in general level of responding. Although incom- plete, the data from further retention testing indicate that the difference be- tweenthesetwogroupspersistsforat least5months.

Affectional retention was also tested SCIENCE, VOL. 130

grees of attractiveness (24). We have utilized this characteristic of response selectivity on the part of the monkey to measure the strength of affectional re- sponsiveness of the babies raised with mother surrogates in an infant version
of the Butler box. The test sequence in- volvesfourrepetitionsofatestbattery CL200 in which the four stimuli of cloth mother, wire mother, infant monkey, andemptyboxarepresentedfora30- WI 100 minuteperiodonsuccessivedays.The

first four subjects raised wth the dual

mother surrogates and the eight infants

raisedwithsinglemothersurrogates
weregivenatestsequenceat40to50 Fig.17.Differentialresponsestovisual

exploration.

days of age, depending upon the avail- ability of the apparatus. The data ob- tained from the three experimental groupsandacomparablecontrolgroup arepresentedinFig.17.Bothgroupsof infantsraisedwithclothmothersshowed Uo,n approximately equal responsiveness to

ovL

VISUAL EXPLORATION RETENTION

DUAL FED I

– CLOTHMOTHER 0-O WIRE MOTHER

— BLANK

the cloth mother and to another infant
monkey, and no greater responsiveness
to the wire mother than to an empty
box.Again,theresultsareindependent 4t ofthekindofmotherthatlactated, W&
cloth or wire. The infants raised with
only a wire mother and those in the
controlgroupweremorehighlyrespon-
sivetothemonkeythantoeitherofthe Fig.18.Retentionofdifferentialvisual-

mothersurrogates.Furthermore,the explorationresponses. former group showed a higher frequency

of response to the empty box than to thewiremother.

Insummary,theexperimentalanaly-
sis of the development of the infant monkey’s attachment to an inanimate mother surrogate demonstrates the over- whelming importance of the variable of
soft body contact that characterized the cloth mother, and this held true for the appearance, development, and mainte- nance oftheinfant-surrogate-mothertie.
The resultsalsoindicatethat,without
the factor of contact comfort, only a
weak attachment, if any, is formed. Finally,probablythemostsurprising 4t

f 1- A

findingisthatnursingor feedingplayed eithernoroleorasubordinaterolein the development of affection as meas-

AL 50c

SINGI- CLOTHNONFED W-_FLOTH MOTHER- 0-4 BLANK

uredbycontacttime,responsivenessto 00WIREMOTHER

fear, responsiveness to strangeness, and motivationtoseekandsee.Noevidence

wasfoundindicatingthatnursingmedi- Fig.19.Retentionofdifferentialvisual- atedthedevelopmentofanyofthese explorationresponsesbysingle-surrogate responses,althoughthereisevidencein- infants.

428

300

U, 200 _

350

3001-

Co W250

CO
WX 200

W 150 W

Z 100I_

21L. uZs

DUAL FED

.iO0

II ORIGINAL0-1

2-3 4-5 RETENTION MONTH

145

X W

CONTROLS

-J10_K >zZz-a– z

SINGLE

SINGLE WIREFED CLOTHFED

LOVE MACHINE MEAN LEVER PRESSES

– SINGLEWIREE.D

–

,O—0 PLANK 995

a 5

ORIGINAL09 30.396069 099 DAYS OF RETENTION

6-9

to9

U) 9 In

°a cl 7

aW 6 t._

5 Z4

3 2

U)5

0 0.

3 p

z2 Wa

__—- — ‘

v – 4 DUAL MOTHERS — SINGLECLOTH

FREE FIELD RETENTION MEAN TIME PERIODS IN CONTACT WITH MOTHER

known. Since the infants raised with bothmotherswerealreadyapproaching themaximumscore inthismeasure, therewas litleroom forimprovement. The infants raised with a single nonlac- tating cloth mother, however, showed a consistent and significant increase in this measure duringthefirst90daysofre- tention.Evidenceforthepersistenceof this responsiveness is given by the fact that after 15 months’ separation from their mothers, the infants that had lived withclothmothersspentan average of 8.75 out of 12 possible time periods in contact with the cloth mother during the test. The incomplete data for reten- tion testing of the infants raised with onlya lactatingwiremotheror a non- lactating cloth mother indicates that thereislitleor no changeintheinitial differences found between these two groups in this test over a period of 5 months. In the absence of the mother, the behavior of the infants raised with clothmotherswassimilarintheinitial retention tests to that during the pre- separation tests, but with repeated test- ing they tended to show gradual adap- tation to the open-field situation and, consequently,a reductionintheiremo- tionality scores. Even with this over-all reduction in emotionality, these infants had consistently lower emotionality scores whenthemotherwas present.

Atthetimeofinitiatingtheretention tests, an additional condition was intro- duced into the open-field test: the sur- rogatemotherwas placedinthecenter oftheroomandcoveredwithaclear Plexiglas box. The animals raised with clothmotherswereinitialydisturbed and frustrated when their efforts to se- cure and contact the mother were blocked by the box. However, after sev- eral violent crashes into the plastic, the animalsadaptedtothesituationand soon usedtheboxas a placeoforienta- tion for exploratory and play behavior.

TIME IN CONTACT WITH MOTHER AA.

*0–_

“0-

o-o DUALFED
*- SINGLECLOTHNONFED 0-0 SINGLE WIRE FED

llAI I ORIGINAL0 30-396069909

DAYS OF RETENTION

Fig. 22. Retention of responsiveness to mother surrogates in the straight-alley test.

Infact,severalinfantswere muchmore active under these conditions than they were whenthemotherwas availablefor direct contact. A comparison of the composite emotionality index of the babiesraisedwithasingleclothorwire mother under the three conditions of no mother,surrogatemother,andsurro- gate-mother-box is presented in Fig. 21. The infants raised with a single cloth mother were consistently less emotional when they could contact the mother but also showed the effects of her visual presence, as theiremotionalityscores in theplasticboxconditionwere definitely lowerthantheirscoreswhenthemother was absent.Itappears thattheinfants gained considerable emotional security from the presence of the mother even thoughcontactwasdenied.

In contrast, the animals raised with onlylactatingwiremothersdidnot showany significantor consistenttrends during these retention sessions other than a general over-all reduction of emo- tionality,whichmay beattributedtoa generaladaptation,theresultofre- peated testing.

Affectional retention has also been measured in the straight-alley test men- tioned earlier. During the preseparation tests itwas found that the infants that hadonlywiremothersdevelopeda gen- eral avoidance response to all of the stimuliinthistestwhen theywere about

100 days of age and made few, if any, responses to the wire mother during the final test sessions. In contrast, al the infantsraisedwitha clothmotherre- sponded positively to her. Maternal separation did not significantly change the behavior of any of the groups. The babiesraisedwithjustwiremothers

Fig. 20. Retention of responsiveness to mothersurrogates intheopen-fieldtests.

intheopenfieldduringthefirst9days after separation and then at 30-day in- tervals. Each test condition was run twice in each retention period. In the initial retention tests the behavior of the infants that had lived with cloth moth- ers differed slightly from that observed during the period preceding separation. Theseinfantstendedtospendmore time incontact with themother and lestime exploringandmanipulatingtheobjects intheroom. Thebehavioroftheinfants raised with single wire mothers, on the otherhand,changedradicallyduringthe firstretentionsessions,andresponsesto the mother surrogate dropped almost to zero.Objectiveevidenceforthesedif- ferencesare giveninFig.20,whichre- vealsthemean numberoftimeperiods spent in contact with the respective mothers. During the first retention test session,theinfantsraisedwithasingle wiremothershowedalmostno responses to the mother surrogate they had always

SINGLE CLOTH NONFED Ul) -*—*CLOTHMOTHER

o-oCLOTHMOTHER-BOX

SINGLE WIRE FED

5 0

U) C/)

cwId

COMPOSITE EMOTIONAL INDEX

IdC4- *—4 BLANK I’4 5t

h3 a

&4 0-=– -~ CL

3 2 –

I –

WIRE MOTHER O—0WIRE MOTHER-BOX

— BLANK

29
A a a _ a A IA I

Fig.21.Emotionalityindexunderthreeconditionsintheopen-fieldretentiontests. while al of the infants raised with

ORIGINAL0-93396990-99

21AUGUST 1959

429

II
ORIGINAL0-930-3960-699099 continuedtofleeintothehidingbooth

O
DAYS OF RETENTION in the presence of the wire mother,

0:4

0-_-

SINGLE WIRE

5 023

RETENTION MONTHS

Fig.23.Typicalresponse toclothmotherinthemodifiedopen-fieldtest.

the infants were forced to approach and bypass the fear stimulus or the wire mother, or both, in order to reach the cloth mother. Following these 24 trials with the mothers present, one trial of each condition with both mothers ab- sent was run, and this in turn was fol- lowed by two trials run under the most emotion-provoking condition: with a me- chanical toy present and the direct path to the mother blocked.

We now have complete data for the firstfourinfantsraisedwithbothacloth and a wire mother. Even with this scanty information, the results are obvious. As would be predicted from our other measures, the emotionality scores for the three stimuli were significantly dif- ferent and these same scores were in- creased greatly when the direct path to the mother was blocked. A highly sig- nificant preference was shown for the cloth mother under both conditions (di- rect and blocked path), although the presence of the block did increase the

cloth mothers continued to respond sponses show resistance to extinction number of first responses to the wire positively to the cloth mother at ap- similarto theresistancepreviouslydem- mother from 3 to 10 percent. In al proximatelythesame levelas inthe onstratedforlearnedfearsandlearned casesthiswasatransientresponseand preseparationtests.Themean number pain.Suchdataare inkeepingwith theinfantssubsequentlyranontothe

oftimeperiodsspentincontactwiththe commonobservationofhumanbehavior.

cause offamiliarizationandadaptation to explore, manipulate, and even attack resultingfromrepeatedtesting.There- and destroy the fear stimuli. It was as fore,attheendof1year ofretention iftheybelievedthattheirmotherwould

It is true, however, that the infants raised with cloth mothers exhibit some absolutedecreaseinresponsivenesswith

cloth mother and clung tightly to her. Objective evidence for this overwhelm- ing preference is indicated in Fig. 24, which shows the mean number of time

appropriate mother surrogates for the
first 3 months of retention testing are
givenin Fig.22.Thereislitle,ifany,
waningofresponsivenesstothecloth timeinalofourmajortestsituations. periodsspentincontactwiththetwo mother during these 3 months. There Such results would be obtained even if mothers. After a number of trials, the appearedtobesome losofresponsive- there were no true decrease in the infantswouldgofirsttotheclothmother nesstothemotherinthissituationafter strengthoftheaffectionalbond,be- andthen,andonlythen,wouldgoout

5to6monthsofseparation,butthetest was discontinuedatthattimeasthein- fantshadoutgrowntheapparatus.

Theretentiondatafromthesemul- testing,new testswere introducedinto protectthem,evenatthecostofherlife tiple tests demonstrate clearly the im- the experimental program. -littleenoughtoaskinviewofhercon- portanceofbodycontactforthefuture Ourfirstnewtestwasamodification dition.

maintenance of affectional responses. of the open-field situation, in which Whereasseveralofthemeasuresinthe basicprinciplesofthehome-cagefear

fectionalbondisformeditismaintained fora very considerablelengthoftime withlitlereinforcementofthecontact- comfortvariable.Thelimiteddataavail- able for infants that have been sepa- ratedfromtheirmothersurrogatesfor a year suggest that these affectional re-

of,theroom directlyinfrontofthestart- box in successive test sessions. Eight trialswere run undereachstimuluscon- dition,andinhalfofthetrialsthemost direct path to the cloth mother was blockedbya largePlexiglasscreen, il- lustrated in Fig. 23. Thus, in these trials

tactswiththeclothmother. Oursecondtestofthisserieswasde-

signedtoreplacethestraight-alleytest described above and provide more quantifiable data on responsiveness to fearstimuli.Thetestwasconductedin an alley 8 feet long and 2 feet wide.

430

SCIENCE, VOL. 130

testwere incorporated.Thisparticular
choicewas madepartlybecausethelat-
ter test had to be discontinued when the
mothersurrogateswere removedfrom Plexiglaspartitionwhichformerlyhad the home cages. blocked their path to the mother, or

preseparationperiodsuggestedthatthe
infants raised with only a wire mother
might have developed a weak attach-
ment toher,alresponsivenessdisap-
peared in the first few days after the
motherwas withdrawnfromtheliving-
cage. Infants that had had the oppor-
tunity of living with a cloth mother
showed the opposite effect and either
becamemore responsivetothecloth placedinthetestroom oppositea testinwhichthemostemotion-evoking mother or continued to respond to her plastic start-box. Three fear stimuli, se- situation was presented resulted in be- at the same level. lected to produce differing degrees of havior near the normal level, as meas-

Thesedataindicatethatonce an af- emotionality,were placedinthecenter uredbytheemotionalityindexandcon-

Forthenew experimenta Masonite theywouldcrouchinthecornerbehind

floor marked off in 6- by 12-inch rec- tangles was placed in the open-field chamber. Both mother surrogates were

the block where the mother normally would have been. The return of the mothers in the final two trials of the

The removal of the mother surrogates fromthesituationproducedthepre- dictableefectofdoublingtheemotion- alityindex.Intheabsenceofthemoth- ers, the infants would often run to the

At one end of the alley and directly behind the monkeys’ restraining cham- ber was a small stimulus chamber which contained a fear object. Each trial was initiated by raising an opaque sliding door which exposed the fear stimulus. Beginning at a point 18 inches from the restraining chamber, the alley was di- vided lengthwise by a partition; this provided the infant with the choice of

entering one of two aleys.
The efects of al mother combina-

TIME IN CONTACT WITH MOTHERS DUAL FED

Z2L W

tightly covered with the same type of cloth. Thus, both objects contain the variable of contact with the soft cloth, but the shape of the mother tends to maximize the clinging variable, while the broad flat shape of the plane tends to minimize it. The preliminary results for differences in responsiveness to the cloth mother and responsiveness to the inclined plane under conditions that produce stres or fear or visual explor- ation suggest that clinging as well as contact is an affectional variable of con- siderableimportance.

tions were measured; these combina-
tions included no mothers, two cloth
mothers, two wire mothers, and a cloth
and a wire mother. All mother con-
ditions were counterbalanced by two
distance conditions-distances of 24
and 78 inches, respectively, from the
restraining chamber. This made it pos-
sible, for example, to provide the in-
fantwiththealternativeofrunningto
theclothmotherwhichwasinclose showedthetypical2:1differentialratio ofinfantsshowasmallbutconsistent proximitytothefearstimulusortothe withrespecttomother-absentand preferencefortherockingobject,as wiremother(ornomother)atagreater mother-presentconditions. measured in average hours spent on the distancefromthefearstimulus.Thus, Theresearchespresentedhereonthe twoobjects. itwaspossibletodistinguishbetween analysisoftwoaffectionalvariables Preliminary results for these three

les,theevidenceisquiteconclusive.A
highly significant preference is shown
fortheclothmotherascomparedtothe
wiremotherortonomother,andthis
preferenceappearstobeindependentof
theproximityofthemothertothefear
stimulus.Intheconditioninwhichtwo
clothmothersarepresent,one24inches tioncoeficients,computedfromfour fromthefearstimulusandtheother78 samplesof100observationsbyfivedif- inchesfromit,therewasapreference ferentpairsofindependentobservers

forthenearestmother,butthedifer- overaperiodofmorethanayear, enceswerenotstatisticalysignificant. rangedfrom.87to.89.
In two conditions in which no cloth
mother was present and the infant had

tochoosebetweenawiremotherand AdditionalVariables
no mother, or between two empty cham-
bers,theemotionalityscoreswerealmost Althoughtheoverwhelmingimpor- twicethoseunderthecloth-mother-pres- tanceofthecontactvariablehasbeen

ent condition. Nodifferenceswerefoundineither

oftheseteststhatwererelatedtoprevi- ousconditionsoffeeding-thatis,to whetherthemonkeyhadnursedonthe cloth or on the wire mother.

Theresultsofthesetwonewtests,in-
troduced after a ful year’s separation
ofmothersurrogateandinfant,arecom-
parabletotheresultsobtainedduring
thepreseparationperiodandtheearly justcontactbutclingingcontact.To retentiontesting.Preferentialresponses testthishypothesis,fourinfantmon- stil favored the cloth as compared to keys are being raised with the standard the wire mother by as much as 85 to cloth mother and a flat inclined plane,

21 AUGUST 1959

431

CLOTH WIRE MOTHER MOTHER

Experiments now in progress on the role of rocking motion in the develop- ment of attachment indicate that this may be a variable of measurable im- portance. One group of infants is being raised on rocking and stationary moth- ers and a second group, on rocking and

Fig. 24. Differential responsiveness in the modifiedopen-fieldtest.

90percent,andtheemotionalityscores stationaryinclinedplanes.Bothgroups

runningtothemothersurrogateasan throughtheuseofobjectiveandob- groupsintheopen-fieldtestgiveaddi- objectofsecurity,andgeneralizedflight servationaltechniquessuggestabroad tionalevidenceconcerningthevariable inresponsetoafearstimulus. newfieldforthestudyofemotional ofclingingcomfort.Thesedatarevealed

Again,thedataavailableatthistime developmentofinfantanimals.The that the infants raised with the stand- arefromthefirstfourinfantsraised analogoussituationsandresultsfound ardclothmotherweremoreresponsive withclothandwiremothers.Neverthe- inobservationsandstudyofhumanin- totheirmothersthantheinfantsraised

sofarobtainedexperimentallyarein excellentconcordancewiththeafec- tionalvariablesnamedbyBowlby(14). Wearenowplanningaseriesofstudies toassesstheefectsofconsistencyand inconsistency with respect to the mother surrogatesinrelationtotheclinicalcon- cept of rejection. The efects of early, intermediate, and late maternal depri- vation and the generalization of the in- fant-surrogateattachmentinsocialde- velopmentarealsobeinginvestigated.

fantsandofsubprimatesdemonstrate withinclinedplanesweretotheplanes. the apparent face validity of our tests. The discovery of three variables of Thereliabilityofourobservationaltech- measurableimportancetotheformation niquesisindicated,forexample,bythe andretentionofaffectionisnotsurpris- correlationcoeficientscomputedforthe ing,anditisreasonabletoassumethat compositeemotionalindexintheopen- otherswillbedemonstrated.Thedata

fieldtest.Fourproduct-momentcorrela-

clearlydemonstratedintheseexperi- Indeed,thestrengthandstabilityof ments,thereisreasontobelievethat themonkeys’affectionalresponsestoa otherfactorsmaycontributetothede- mothersurrogatearesuchthatitshould

be practical to determine the neurolog- pattern.Wearecurrentlyconductinga icalandbiochemicalvariablesthatun-

derlie love.
References and Notes

1. K.Lorenz,Auk54,245(1937).
2. R. A. Hinde, W. H. Thorpe, M. A. Vince,

Behaviour9,214(1956).
3. E.Fabricius,ActaZool.Fennica68,1(1951). 4.E.H.Hess,1.Comp.andPhysiol.Psychol.,

inpress.
5. J.Jaynes,ibid.,inpress.
6. H. Moltz and L. Rosenblum, ibid. 51, 658

(1958).

velopmentoftheaffectionalresponse

series of new experiments to test some ofthesepostulatedvariables.

For example, Bowlby (14) has sug- gested that one of the basic affectional variablesintheprimateorderisnot

7. C. R. Carpenter, Comp. Psychol. Monograph No. 10 (1934), p. 1.

8. A. Nolte, Z. Tierpsychol. 12, 77 (1955).

9. S. Zuckerman, Functional Afinities of Man, Monkeys and Apes (Harcourt Brace, London,
1933).

10. W. Kohler, The Mentality ofApes (Humani-
tiesPress, New York, 1951).

11. R. M. Yerkes and M. I. Tomilin, 1. Comp.
Psychol. 20, 321 (1935).

12. J. Dollard and N. E. Miller, Personality and

Psychotherapy (McGraw-Hill, New York, 1950), p. 133; P. H. Mussen and J. J. Con- ger, Child Development and Personality (Harper, New York, 1956), pp. 137, 138.

13. M. A. Ribble, The Rights of Infants (Co-

lumbia Univ. Press, New York, 1943); D. W. Winnicott, Brit. J. Med. Psychol. 21, 229 (1948).

14. J. Bowlby, Intern. J. Psycho-Analysis 39, part 5 (1958).

15. Support for the research presented in this article was provided through funds received from the graduate school of the University of Wisconsin; from grant M-772, National Insti- tutes of Health; and from a Ford Foundation grant.

16. G. van Wagenen, in The Care and Breeding of Laboratory Animals, E. J. Farris, Ed. (Wiley,NewYork,1950),p.1.

17. J. P. Foley, Jr., J. Genet. Psychol. 45, 39 (1934).

18. We no longer make the cloth mother out of a block of wood. The cloth mother’s body is simply that of the wire mother, covered by a tery-clothsheath.

19. H. F. Harlow, Am. Psychologist 13, 673 (1958); and R. R. Zimmermann, Proc. Am. Phil. Soc. 102, 501 (1958).

20. D. 0. Hebb, The Organization of Behavior (Wiley,New York,1949),p.241f.

21. A. T. Jersild and F. B. Holmes, Child De- velop. Monograph No. 20 (1935), p. 356.
22. J. M. Arsenian, J. Abnormal Social Psychol.

38,225 (1943).
23. R.A.Butler,1.Comp.9ndPhysiol.Psychol.

46,95 (1953).
24. , J. Exptl. Psychol. 48, 19 (1954).

quired at least 13 amino acids for sur- vival and growth. Over and above the eightamino acidsrequiredfornitrogen balance, these cell cultures require argi- nine, cyst(e)ine, glutamine, histidine, and tyrosine, and on the omission of any one of these, the cels degenerate and die.Intheirearlystagesthesedegenera- tive processes are reversible. This pre- sents an opportunity to explore the inti- mate structural derangement caused by specific amino acid deficiences, and the reparative processes which occur on res- toration of these amino acids to the medium.

None of the D-amino acids substitute for the L-isomer (8) except for D-cystine, and this apparently acts by mobilizing cyst(e)ine residues bound to the serum protein of the medium. (This is dis- cussed below.) Dipeptides were found to be active, substituting for both com- ponent amino acids (10). Recent ex- perimentsindicatethat,althoughdipep- tides are hydrolysed extracellularly by serum peptidases, and perhaps by cel- derived peptidases, some of the dipep- tideistransportedintothecellandthere hydrolyzed.

A number of keto acids and other amino acid congeners have been tested withrespecttotheirabilitytosubstitute for the corresponding amino acid (11). The results with several human and ani- mal cell lines are summarized in Table 2. The cels contain a wide variety of transaminases (12), and, as determined in feeding experiments in rats (5), most of these keto acids do in fact substitute for the corresponding amino acid.

Of particular interest is the fact that, with al the human cell lines studied, citrulline substituted for arginine, while ornithine was inactive [Table 2; see also

(13)]. Attempts to encourage citrulline biosynthesis by progressive removal of

– Theauthorischiefofthelaboratoryofcel biology,NationalInstituteofAllergyandIn- fectious Diseases, National Institutes of Health, Bethesda,Md.

Amino Acid Metabolism in Mammalian CellCultures

Harry Eagle

Anumberofcellinesderivedfrom suspensionculturesindicatesthatthisis normal and malignant tisueshave now not itsonly function, and itseems rea- beenserialypropagatedinculture. sonabletoassumethatitactsinpartas Thesemaybegrownadherenttoaglas acarrierofasyetunidentifiedgrowth surface and overlaid with a fluid me- factors which are bound to the protein, dium,ortheymaybegrowninsuspen- andwhichareslowlyreleasedintothe sion;butineithercasetheopportunity medium. ispresentedforthestudyofmetabolism The presentarticle,dealingwiththe atthecellularlevelinasystemwhich aminoacidmetabolismofthesecultured operationallyresemblesbacterialcul- cels,isaprogressreportratherthana turesinmostessentialrespects.Thecels reviewandinlargepartsummarizes and the medium can be separately ana- studies from a single laboratory. The lyzed,balanceexperimentscanbesetup, enormousbodyofinformationavailable metabolicprocessescanbeexamined withrespecttoaminoacidmetabolism

qualitativelyandquantitativelyunder inbacterialcultures(5,6)hasservedas controlledconditions,andthecorre- astimulusandprototypeforthestudies sponding enzymatic activities can be ex- here reported. The relevant findings in plored in cel-fre extracts. bacterial systems have not been referred

Arelativelylimitednumberofmetab- toindetail,onlybecausethatimportant oliteshavebeenshowntosuficeforthe exerciseincomparativebiochemistryis apparentlyindefinitepropagationofal beyondthescopeofthepresentreport. thehuman cellinessofarstudied.The

minimal medium, in which every com- ponentisdemonstrablyesential,islisted NutritionallyEssentialAminoAcids in Table 1 and includes 29 components:

13 amino acids, 8 vitamins, 6 ionic spe- The classic experiments of Rose and cies,glucose,andserumprotein(1,2). hiscoworkers(7)haveshownthateight The role of the serum protein is not en- amino acids (isoleucine, leucine, lysine, tirelyclear.Althoughafewcellines methionine, phenylalanine, threonine, havebeenserialypropagatedinapro- tryptophan,andvaline)suficefornitro- tein-freemedium(3),serumproteinis genbalanceinfeedingexperimentsin requiredbymostmammaliancellcul- man.Incontrast,-everycelcultureso tures.Inmonolayercultures,theprotein farexamined,whetherhumanoranimal playsaroleintheadhesionofthecels (8)inorigin,andwhetherderivingfrom toglas(4).However,itsesentialityin normalormalignanttissue(9),hasre-

432

SCIENCE, VOL. 130