You are on page 1of 49

Society for Research in Child Development

Wiley

The Modification of Social Responsiveness in Institutional Babies


Author(s): Harriet Lange Rheingold
Source: Monographs of the Society for Research in Child Development, Vol. 21, No. 2, The
Modification of Social Responsiveness in Institutional Babies (1956), pp. 1-48
Published by: Wiley on behalf of the Society for Research in Child Development
Stable URL: http://www.jstor.org/stable/1165614
Accessed: 17-10-2015 10:46 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/1165614?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Society for Research in Child Development and Wiley are collaborating with JSTOR to digitize, preserve and extend access to
Monographs of the Society for Research in Child Development.

http://www.jstor.org

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
ofthe
Monographs
SOCIETY FOR RESEARCH IN CHILD DEVELOPMENT, INC.

VOLUME XXI * SERIAL NO. 63 * No. 2 * 1956

ofSocial
The Modification
Responsiveness
in Institutional
Babies

HARRIET LANGE RHEINGOLD

NationalInstituteofMentalHealth
NationalInstitutesof Health
Public Health Service
U.S. Departmentof Health, Educationand Welfare

CHILD DEVELOPMENT PUBLICATIONS


1956

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
EDITOR
E. MARTIN
WILLIAM
PurdueUniversity

MANAGING EDITOR
KATE HOFFMAN

Publishedby

CHILD DEVELOPMENT PUBLICATIONS


of theSocietyforResearchin ChildDevelopment,
Inc.
PurdueUniversity
Indiana
Lafayette,

1956bytheSocietyforResearchin ChildDevelopment,
Copyright Inc.

Printed
by
THE ANTIOGH PRESS
YellowSprings,
Ohio

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
CONTENTS

Page
THE PROBLEM 5
................................................
EXPERIMENTAL ........................................
PROCEDURE 8
The Institution.......................................... 8
The Subjects ............................................ 9
Experimental Treatment.................................. Io
Time-Sampling Measuresof Treatment ..................... 11
The Tests 14
...............................................
RelationsbetweenTestData and thePropositions ............ 18

THE RESULTS...................................................... 19
ObservationalData ....................................... 19
Comments on Treatment oftheData ....................... 20
Noteon Presentation of theData ........................... 21
The SocialTest .......................................... 21
TestofPosturalDevelopment ............................. 36
Testof Cube Manipulation ............................... 38
The CattellInfantIntelligence Scale ....................... 38

DiscussIoN ............................................................. 4

SUMMARY
........................................................ 44
REFERENCES
.................................................... 47

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
ACKNOWLEDGMENTS

This reportis based upon a dissertation submitted in


1955 to theDepartment of Psychology at theUniversity
of the requirements
of Chicagoin partialfulfillment for
the degreeof Doctorof Philosophy.The writertakes
pleasurein acknowledging the help receivedfromDr.
Helen L. Koch, sponsorof the dissertation; fromDr.
Donald W. Fiskeand Dr. AustinH. Riesen,members of
thecommittee; fromDr. JacobL. Gewirtz,Dr. Howard
F. Hunt, and Dr. Lyle Jonesof the Department; and
fromSisterAntoniaand SisterAnthony of St. Vincent's
Infantand Maternity Hospitalwherethestudywas per-
formed. During the period, 1954-1955, the author was
a PublicHealthFellowin ClinicalPsychology.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE PROBLEM

This is an attemptto explorethe effectupon infantsof an experimental


modificationof the environment.For a group of institutionalbabies one
"mother"was substitutedfor many "mothers"and the effectupon social
responsivenesswas assessed. This was the primaryfocus of the study,but
the effectsupon other kinds of behavior were also measured. The main
question asked was, "What effectwill an increasein social responsiveness
to one person,the 'mother,'have upon responsiveness to otherpersons?"
From almost the beginning of life the infantis interestedin people.
His firstactive social response has been defined by Shirley (25) as the
sober watchingof an adult face early in the firstmonth of life. At five
weeks of age he is already smiling in response to an adult's overtures
(6, 24). With time his social responsivenessbecomes more elaborate; he
laughs, vocalizes,stretchesout his hands to be picked up, grasps another's
clothes (5, 23)-
Althoughat firsthe is interestedin and responsiveto all people, in the
second monthhe discriminateshis motherfromthe othersabout him and
shows his recognitionby change in expression,an intentnessof regard,
and by more ready and fullersmiling (18). In anothermonth he sobers
at the sight of a stranger (1I, 15, 25). Facial expressions of distress and
fear,crying,and physicalwithdrawalcharacterizelaterresponsesto strang-
ers. Bayley (2) cites "discriminatesstrangers"as a normativeitem at 5.5
months which she defines as staring,frowning,withdrawal,or crying.
Crying in responseto strangersin an examinationsituationincreasesup
to io monthsof age (i). Gesell and Thompson (15) point to the period
between 28 and 32 weeks for the change from "often discriminates
strangersbut usually adapts well to them" to "likely to withdraw from
strangers."Hebb and Riesen (20) have describedsimilar behavior in the
young chimpanzee.
In social behavior,as in otherkinds of behavior,the normativedeline-
ation of developmentobscures individual differences. For example, some
babies appear never to be shy of strangers(25) and percentagesreported
by Gesell and Thompson (I5) for withdrawingfrom strangersdo not
reach 50 at any age level. The causes of these differences
have been attrib-
uted bothto natureand to nurture.Social behavior,however,has been con-
sidered especiallysusceptibleto environmentalevents (04, 25). It is the
effectof environmentaleventswhich we will pursue here.
Institutionswith theirmultiple"mothers,"less frequentcontactwith
adults, more impersonalcare, provide an environmentwhich differsmark-
edly from familylife and, as might be expected,the social behavior of
institutionalizedbabies departs from normal development.These babies
have been reportedto be less interestedin people, less responsiveto them,
5

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
6 MODIFICATION OF SOCIAL RESPONSIVENESS

less sensitiveto changes in an adult's facial expressionand tone of voice


(13, 14, 26). That theyalso show less interestin thingsand less competence
in manipulating them has been attributedto this departure from the
mother-child interplayof familylife,ratherthan to the absence of material
thingsin theirenvironment.
For familybabies,littleis definitelyknown about environmentalevents
which may affecttheirsocial responsivenessto known and strangepersons
or about the effectof social experienceupon the manipulationof things.
Normal social responsivenessdid develop in the twins reared by the Den-
nises (Io) under conditionsof "minimum" social stimulation,although
in a reprintingof the study(12) the experimentaltreatmentwas described
as "controlledenvironmentalconditions,"conditions which Stone (28)
believes qualify as "minimum adequate social stimulation."This experi-
ment excepted,we have only opinion, such as Hebb's (19), based upon
"common experience" that the fear of a strangerwill be minimized or
absent if a baby has already been exposed to the sight of a large number
of persons; Jersild'ssuggestion(2I) that awareness of a strangerwill be
affectedby the experiencesthat babies have had with people; and Bayley's
observationthat,"When a child is old enough to make such differentiations,
strange persons and places-if he is not taken about frequently-are a
definitestimulusfor crying" (I, p. 316).
For this experimentinfantsliving in an institutionwere chosen as
subjectsbecause of their common backgroundof experienceand because
of the relativeease with which environmentalconditionscould be con-
trolled and manipulated. We decided, also, to work with six-month-old
infantsbecause, by this age, familybabies oftengive clear evidence of a
difference in theirresponsesto known and unknown persons.
Againstthisbackgroundtheproblemwas set.The questionsasked were:
I. Can six-month-old infantswho have been cared for in an institution
by many different personslearn to respondselectivelyto one person who
assumes theircare fora period?
2. If the infantslearn this discrimination,how will it affecttheir re-
sponses to other persons? Specifically,will it produce an awareness of
strangers?Will this awareness be characterizedby signs of discomfort,
fear,or physicalwithdrawal?
3. What will be the effectof care by one "mother" upon their
achievementsin certain other areas of behavior?
To answer these questions a series of propositionsupon which experi-
ment and observationmight be expected to yield informationwas set
forth,as follows:
i. Infantsof six monthsof age who have been cared for by multiple
"mothers,"varyingfromday to day, should award special recognitionto
no one person.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE PROBLEM 7
2. These infants shouldshowno awarenessof a stranger.
3. If,now,one person,in contrast to manypersons, assumestheircare
fora periodof time,thereshouldbe an increasein socialresponsiveness
to thatperson.
4. Followingthe appearancein the child of a differentiatedresponse
to themother figure, thereshouldappearan awareness resem-
of strangers,
blingin sequenceand patterning of responses thecourseof development
of normalbabies,includingresponses of discomfort
and withdrawal.
5. Finally,care by one personshouldresultin improvedperformance
in otherareasof behavior.Posturalachievements, successin manipulating
cubes, and performance on a testof werechosenas tests
"intelligence"
of development in otherareasof behavior.
The primary focusof thisstudywas theinfant'ssocialbehavior;other
areaswereconsidered as secondaryto thepurpose.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
EXPERIMENTAL PROCEDURE

THE INSTITUTION
The institutionwas a large urban maternityhospitaland orphanasylum
whichcared forchildrenfrombirththroughthreeyearsof age. The reasons
for the children'sresidencein the institutionwere varied. They included
such causes as abandonment,financialinabilityof the parentsto care for
the child, and illnessof the mother.Physicalplant and care were excellent.
The institutionobservedthe usual hospital practicesto minimize the
spread of infection.Wearing a gown (masks were not used) and scrubbing
between handling babies were included in these practices.For the same
reason,babies could not be put in a play pen or on the floor,or one baby
broughtinto contactwith another.
The babies were cared for by studentsin a one-yearchild-carecourse,
by supervisorswho were graduatesof the course,and by volunteers.The
volunteerswere women fromthe communitywho came for a few hours
of the day, usually from 9 A.M. to 12:30 P.M. During this period they
were the ones who fed, bathed, and generallycared for the babies. The
volunteerscame on only a personally-fixed schedule,usually once a week.
No volunteercame everyday or even everyotherday. The numberof vol-
unteersvaried fromday to day, and fromweek to week. In theirabsence
the studentsand theirsupervisorscared for the babies. Thus a baby would
ordinarilynot be cared forby the same personfromone time to the next.
As a rule, the caretakers,whether volunteersor hospital personnel,
worked togetheras a group,caring for the babies in turnaccordingto the
positionof theircribs in the room, and upon completingthe task of the
moment,bathing,or feeding,or diapering,moved on as a group to the
next room.
The routineof a typicalday follows: Breakfastwas broughtto the
floorat 7:30 A.M.Only a rarevolunteerarrivedthis early,so thatit was the
studentnursesand theirsupervisorswho made the firstcontactswith the
babies. Each child was diapered,fed, and then returnedto his crib before
the next child was attendedto. For his meals the child was held in arms
or put in a high chair. Usually only two or three people were available
to feed the 16 babies in a ward so that all were not fed until 9 A.M. It
was about this time that the volunteersarrived.The next major routine
was bathing,dressing,changing of sheets,and sterilizingthe beds. Every
otherday the baby was taken out of his bed to be bathed and dressed.On
alternatedays he was sponge-bathedand dressedin his bed. The volunteers
did almost all the bathingand dressing,hospital personnelcaring for the
beds. This routinewas completedonly just before :i A.M. when food for
the next meal was deliveredto the floor.Again the babies were changed
and fed,an operationwhich lasted until 12:30 P.M. At this timethe volun-
8

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
EXPERIMENTAL PROCEDURE 9
teers left. An occasional volunteerreturnedor came in afterlunch. She
diapered,played with,and generallycared forthe babies in the threerooms
which comprisedone ward. On occasion there was no one at all in the
rooms during the early afternoon.The studentnurses were not on duty
and the supervisorsworkedon theircharts.Certainlyduringthistime there
were no major caretakingactivities.At 3 P.M. a volunteeror two would
be presentto assist the studentnurses in feeding the children the third
meal of the day, and mightstaythroughthe late afternoon.
The detailsof the hoursfrom3 P.M. to 7 A.M.the writerdoes not know
at firsthand. By reportdiapers were changed periodically,the babies were
dressed in freshclothesfor the night about 6 P.M., and two bottleswere
given,one in the eveningand one duringthe night.Most of these services
were performedby the studentnurses.
This recitalof activitieswould be incompletewithouta word about the
manner in which the daily routineswere carried out. The characteristic
attitudeof all caretakers,whethervolunteersor hospital personnel,was
benevolent.A majorityof the caretakerswere affectionateand playfulin
caring for the babies. The Sister in charge of the floorencouraged care-
takersto talk to the babies and to hold them.Thus most contactsbetween
child and adult were pleasant.Departuresfromwhat is thoughtto be good
motheringoccurredmore oftenbecause of insufficient personnelthan be-
cause of emotional resistanceor ignorance.
This sketchof the institutiondescribesthe environmentof all the sub-
jects beforethe experimentbegan. It describesalso the environmentfor
the controlbabies during the study.

THE SUBJECTS
The subjects were 16 infants,divided equally into experimentaland
controlgroups.Four pairs of experimentaland controlbabies were studied
in the firstexperiment;the otherfourpairs constituteda second and paral-
lel experiment.
The design called for infants who were six months old, full-term,
withoutmarkedphysicalor mentaldefect,who had lived in the institution
for at least threemonths.An additional requirementwas that therebe a
reasonablygood likelihood that the subjectswould remain in the hospital
for the durationof the experiment.These requirementswere not easy to
achieve.
From the available babies in the firstexperimentwe excluded one baby
who was spasticand two apparentlynormal babies because theirdeparture
from the institutionwas imminent.To obtain the eight babies it was
necessaryto include one baby in his fifthmonthand one who had a Cattell
IQ of 75. For the second experimentwe excluded one blind baby, one
who had a CattellIQ of 5o, and one who was expectedto leave the institu-
tion shortly.Once again it was difficult
to obtaineightsubjects;fourinfants

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
10 MODIFICATIONOF SOCIAL RESPONSIVENESS
in theirfifthmonthof lifeand two in theirseventh, and againone who
had an IQ of 75 had to be used.In otherwords,we werenotable to select
thesubjectsout of a largernumberwho mettherequirements, butrather
had to takethosewhocameclosestto fulfilling a setof specifications.
The eightsubjectsof each experiment weredividedintofourmatched
pairs on thebasis of threevariables,
age, CattellIQ, and socialresponsive-
ness.Becausethesamplewas small,matching was perforce rough.
In thefirstexperiment theassigning of subjectsto experimentalor con-
trolconditions was restricted
byhospitalarrangements overwhichwe had
littlecontrol.
For thesecondexperiment, however, thechoiceofthemember
in eachpairto receivetheexperimental treatment was madeat random.
Table i summarizes the data forthe originalpopulations of bothex-
periments on thevariablesused formatching, as wellas givingthelength
oftimethesubjects hadbeenin theinstitutionat thebeginning ofthestudy.

TABLE I
MEANS AND RANGES OF VARIABLES USED FOR MATCHING
(N equals 8 in each group)

Experimental
Group ControlGroup
Item Mean Range Mean Range
Chronologicalage ........... 6.1 5.1-7.3 6.2 5.2-7.4
Cattell IQ ................. 96.6 75-119 94.2 75-108
Social score ................ 27.2 12- 43 28.4 17- 41
Lengthof timeinin institution*.. 5.1 3.2-7.3 5.6 3.4-6.8

* Lengthof timein institution


not used as variablein matching.

Four babiesleftthe institution duringthecourseof the experimental


treatment. Two experimental babies,onein eachexperiment,
leftat theend
ofthethirdweek.Two controlbabiesofthesecondexperiment leftduring
theseventhweek.Aftertheend of theexperimental treatmenttherewere
stillotherlosses.No selective
factorwhichmightaffecttheresultsappeared
tohavebeenresponsible fortheremovalofthesebabiesfromtheinstitution.
Sex of subjectwas notusedas a variablein matching
or in theselecting
of subjects.In thefirstexperiment the experimental
groupwas composed
of two boysand twogirls;thecontrolgroup,of one boyand threegirls.
In thesecondexperiment thereweretwoboysand twogirlsin eachgroup.

EXPERIMENTAL TREATMENT

The experimentercaredfor,"mothered," the foursubjectswho were


assignedto the experimental
conditionin each experiment.These babies
livedin one roomwiththeircribssidebysidealongone wall of theroom.
The experimenter fed, bathed,diapered,soothed,held, talked to, and

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
EXPERIMENTALPROCEDURE 11
playedwiththesefourbabiesforsevenand one-halfhoursa day,from
to five days a week, for eight weeks, a total of 300
3:oo P.M.,
7:30 A.M.
hours.Duringthesehoursno one else caredforthesebabies,althoughat
timesotherpeoplewerein theroom,thenurse,thedoctor,visitors, or the
studentnurseswho made the beds.After3 P.M.the experimental babies
werecaredforaccordingto the hospitalroutinedescribedearlierunder
Institution.
The same routinewas followedon Saturdaysand Sundays,
exceptthattheexperimenter was presenton Saturdays fortheteststo be
describedshortly.
The experimenter and consistently
deliberately triedto adapthercare
to theindividualneedsof eachbabyas thesewereapparentto her,limited
onlyby hospitalroutinesand by thedemandsof caringforfourbabiesat
once.Everyeffort was made to preventa child'scryingor to placatehim
ifhe did cry.The babieswereoutof theircribsoften,beingheldor seated
in chairs.The experimenterput toysin theirhandsand playedwiththe
children.She heldeachchildon herlap forsomeperiodof everyday.She
smiledand talkedto thebabies,and triedto missno opportunity
to respond
to theirsmilesandvocalizations.
The goalwas togivethechildren maximal
gratification.
Duringthe experimentalperiodof eightweeks,the experimenter had
no contactof any kind withthe controlsubjects,exceptforthe weekly
testingperiods.These childrenhad theircribsin anotherroomand were
cared forby hospitalpersonneland volunteers accordingto the routine
describedunder Institution.
At theendoftheeightweeks,theexperimental
subjectsreturned
to the
full-time
careof thehospital,
and theexperimenter
saw themonlyat the
foursubsequentweeklytestingperiods.

TIME-SAMPLING MEASURES OF TREATMENT

To obtaina measureof the differencebetweenthe experimental and


thecontrolconditions,
a separateperson,an observer,
tooktimesamplesof
thenumberof adultsin eachroomand of theiractivities
in relationto the
babies. These variableswere recordedon one half day (7:30 A.M. to noon
or noon to 3:00 P.M.) of each week of the experimentaltreatment.
Further-
more,to ruleout variations
in numberof volunteers
presenton any one
day of the week,different
daysof the week weresampled.The observer
standingin the hall made her observations
throughthe glass windows
betweenthe corridorand the rooms. She recordedfor ro-minuteperiods,
betweenexperimental
alternating and controlroomssystematically
so that
overtheeightweeksofeachexperiment
all possiblero-minute
periodswere
coveredforeach room.
At the beginningof a Io-minuteperiod, the observerglanced briefly
at thebabyin thefarthest
crib (Baby I) and recordedobservations
on a

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
12 MODIFICATIONOF SOCIAL RESPONSIVENESS
checklist. This constituted one observation. Then Baby2, nextnearerher,
was observed, thenBaby3, Baby 4, Baby i again,and so on in rotation
untiltheend of the io minutes.
Beforethe beginningof each experiment, bothexperimenter and ob-
servermade independent simultaneous observationsuntil theyreached
at least90 percentagreement on itemscheckedduringa Io-minute period.
Thereafter, similarstudiesof reliability weremade duringthe courseof
eachexperiment. The percentage of agreement on itemscheckedwas high.
Duringthefirstexperiment threeperiodsof independent observationgave
percentages of 97, 99, and 99; duringthe secondexperiment fiveperiods
gavepercentages of 98, 98, 98, 98, and ioo.
During the two experiments, 168 io-minuteperiodswere sampled.
These yielded7,Ii6 observations, approximately 445 observations per
subject.
Table 2 showsthattherewas considerable resemblance betweenthe
two experiments in thevariablesmeasuredby the time-sampling. In both
experiments thecontrolbabieswerealone,withoutan adultin theroom,
on morethanhalfof theobservations (64 percentin thefirstexperiment,
74 percentin thesecondexperiment). In contrast,
at leastone person,the
experimenter, was in theexperimental roomfor60 percentof theobserva-
tions.In thefirstexperiment caretaking actswererecorded for23 percent
of theobservations on the experimental room,all of themperformed by
the experimenter, whilein the controlroomcaretaking acts totaledonly
7 percentoftheobservations, and theseactsweredividedamong14 differ-
ent persons.In the secondexperiment caretakingacts wererecordedfor
19 per centof the observations on the experimental room,all performed
by theexperimenter, withcaretaking actsin the controlroomappearing
again in only7 per centof theobservations, thistimedividedamong21
different persons.(It shouldbe notedthatthenumberof possibleobserva-
tionsforcaretaking acts in the experimental roomcould not exceed25
per cent becausein any one round of observationson thefourbabiesthe
experimenter could not be caring for more thanone baby.In thecontrol
room,withits manycaretakers workingsimultaneously, thispercentage
could have been exceeded.)The resultsshow also thatthe experimental
childrenwerein theircribson six out of io observations, thecontrolchil-
dren,on nineout of io observations.
Becauseeachseparateobservation was notmadeat a fixedtimeinterval
butinsteadas manyobservations as possibleweremade in the Io-minute
period, unequivocalstatements about percentagesof time cannot be
made. Inspectionof the recordsindicatesthat rate of observing, and
henceof recording, was slowerwhen therewas moreactivity to record,
thatis, eitherwhentherewerepeoplein the roomor whenthesepeople
weredoingsomething forthebabies.Fromthisfactwe concludethatmore,

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
EXPERIMENTAL PROCEDURE 13

TABLE 2
TIME-SAMPLING
OBSERVATIONSOF ACTIVITIESOF CARETAKERSAND
BABIES IN THE EXPERIMENTAL AND CONTROL ROOMS

EXPERIMENT 1 EXPERIMENT 2
E C E C
ItNcm N % N % N % N %
ofobservation
Periods ...... 34 - 34 - 50 - 50
Observations
............. 945 - 1131 - 2233 - 2807
Numberof adultsin room:
0 ................... 212 22 728 64 574 28 2084 74
1 ................... 565 60 262 23 1371 61 381 14
2................ 145 15 88 8 233 10 214 8
3 or more........... 23 2 53 5 55 2 128 5
Caretakers............... 1 -- 14 -- 1 - 21 -
Caretaking acts ........... 215 23 83 7 435 19 185 7
Natureofact:
talking.............. 119 13 25 2 251 11 76 3
caringfor ............ 105 11 34 3 138 6 65 2
playingwith ......... 28 3 5 0 33 3 13 0
holdingin arms ...... 34 4 2 0 112 5 37 1
feeding.............. 37 4 40 4 114 5 44 2
Child awake ............. 829 88 791 70 1752 79 2146 76
awakeand crying 17 2 11 1 43 2 101 4
asleep ............... 116 12 340 30 481 21 661 24
Childin crib............. 567 60 1062 94 1401 63 2535 90
outof crib ........... 120 13 58 5 339 15 108 4
seatedin chair........ 258 27 11 1 493 22 152 5
actscounted
NoTE.-Caretaking onlyonceperobservation;
nature
ofactsnotmutually
exclusive.

ratherthan fewer,observationsof caretakingacts would have been recorded


in the experimentalroom if the rate of observationhad been constant.
Additional informationwas obtained by posting,from time to time,
at the footof the baby's crib a formupon which each adult caring for the
child wrote her name. The average controlbaby was cared for by 13 dif-
ferentpersons in a 24-hour day, the average experimentalbaby by six,
one of whom was the experimenter, the othersperformingtheir services
afterthe experimenterleftfor the day.
These observationsand recordsdemonstratethat the experimentaltreat-
ment did produce an environmentdifferent from the usual environment
of the institution.The experimentalchildrenwere less oftenalone, they
were cared for more often,they were more oftenout of theircribs,they
receivedmore of theircare from one person alone, and they were cared
forby fewerpersons.
To remove any possible advantage accruing to position of child's crib
in the room, the cribs of both the experimentaland controlsubjectswere
changed systematically once a week.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
14 MODIFICATIONOF SOCIAL RESPONSIVENESS

THE TESTS
The experimenter gavenoneof thetestsbecauseit was anticipated that
she mightacquirecertainbiasesbyherknowledge of thehypotheses to be
testedand byherintimate contactwiththeexperimental subjects.
All teststhroughout bothexperiments were administered by another
person(not the observer),hereafter called the examiner.The examiner
was a graduatestudent, trainedby the experimenter, whomwe triedto
in
keep ignorance of the experimental design.It is certainthatshe did
notknowthedimensions of theexperimental condition or thehypotheses
to be tested,but at theend of theninthweekof thefirstexperiment she
reported thatshe knewwhichwas theexperimental room.For thesecond
experiment we tooktwoprecautions to precludethepossibility of a similar
discovery. First,the examinersuspendedstudiesshe was makingof the
"autistic"movements1 of thesebabieswhenthe experimenter was not in
thehospitaland,second,somesubjectsweremovedfromtheexperimental
to thecontrol room,and viceversa,beforeeachtestperiodso thattheposi-
tionofanysubjectin eitherexperimental or controlroomvariedfromweek
to week forthe durationof the testingperiod.Since the positionsof all
cribswithinbothroomswerechangedweekly,thisadditionalalternation
of somebabiesbetweenroomsservedto makediscovery The fail-
difficult.
ure to keep theexaminerin completeignoranceis a regretted departure
fromthedesignof theexperiment.
The batteryoftestsusedwas composedoftheCattellInfantIntelligence
Scale,a testof socialresponsiveness,one of posturaldevelopment, and one
of cubemanipulation. The testsweregivento all thesubjectsin theweek
precedingthebeginning of the experiment, beforeassignment of subjects
to thetreatment condition.The Cattelltestwasgivenonlyonceagain,in the
seventhweek.All othertestswereadministered to bothexperimental and
controlsubjectsbiweekly, on Saturdays, as follows:one partof the social
testand thetestof posturaldevelopment weregivenat theend of thefirst
week of treatment and on everyotherSaturdaythereafter throughthe
eleventh week (i.e., thethirdweekaftertheend of treatment); theother
partof thesocialtestand thetestof cubemanipulation weregivenat the
endofthesecondweekoftreatment and on alternating Saturdays thereafter
through thetwelfth week (the fourthweekaftertheend of treatment).
Testingwas carriedout earlyin themorning, usuallydirectly afterthe
babywas fed.
The SocialTests
Thereweresevensocialtests,thefirst
fourcomposing
PartA, thesecond
three,PartB. These twopartswerenotequated;thesecondpartwas em-
1 The examiner,Ann Baer, recorded
frequenciesof such activitiesas finger-sucking
and rockingmovementsby time-sampling.This studyis in preparation.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
EXPERIMENTALPROCEDURE 15
ployedto yieldadditional
information.
Hereafter
eachseparate
testwill
becalleda subtest.
Theydonotcorrespond
totheordinary
"personal-social"
itemsof normativeschedules,with the exceptionof Biihler's(5, 7), to
whichtheyowe a greatdeal.
The foursituations
forPart A were:
I. The adultstoodapproximately
threefeetfromthe child'scrib,
smiledat thechild,butdid notspeak.
2. The adultwentto thecrib,leanedoverthechild,smiled,and said
warmly, "Hello,baby,howareyou?,"whichcouldbe repeated a second
time.
3. The adulttriedbyanymeanstogetthechildto smile, andas soon
as thechildsmiled,frowned and scoldedhimin an angrytoneofvoice,
saying,"Younaughty baby,whatdidyoudo?" Thesewordstoocouldbe
but
repeated only once.
4. Afterthelapseofat least15minutes(in orderthatthebabymight
forgetthescolding),theadult,concealed
behinda sheetthrown overthe
footofthecrib,calledto thechild,saying,"Hello,baby,"or "Comeon,
baby."
The threesituations forPartB were:
i. The adultstoodat thesideof thecrib,smiled, and talkedto the
childas in A, 2.
2. The adultpickedthechildup, walkedto thecenter of theroom,
heldhimin herarmsso thatthetopof his headwas levelwithher
shoulder,smiled, andtalkedtohimas in A, 2.
3. Theadultreturned thebabytohiscrib,placedhimonhisback,and
stoodat hiscrib,silent, andnotlooking at him.
The adults,or stimulus to whomthebabies'responses
objects, were
obtainedwere(a) theexperimenter, (b) theexaminer, and(c) a volunteer.
The experimenter was a knownpersonto theexperimental babiesbut
unknown to thecontrol subjects.The examiner wasequallyunknown to
bothgroups ofsubjects. Becausethesetestsweregivenso often, however,
thepossibility
existed thatbothexperimenter andexaminer might in time
become knownto all subjects. Thiswasespecially truefortheexaminer
whohadmorecontact withthecontrol thandidtheexperimenter.
subjects
Forthisreason, a volunteer whohad no priorcontact withanyof the
subjects
approached all thebabiesin bothexperiments in theeighth week
onlyandthenonlyforsubtests I and2 ofPartA.
Wheneither theexperimenter orthevolunteer wasthestimulus object,
theexaminer waspresent, althoughas inconspicuousas possible,
to record
thebabies'responses. Butwhentheexaminer wasthestimulus the
object,
experimenterwasnotpresent, and theexaminer recordedtheirresponses
byherself.
To insuretheexaminer's theexperimenter
objectivity, triedin every
waypossibletoapproach all subjects,
experimentalandcontrol,in a similar

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
16 MODIFICATION OF SOCIAL RESPONSIVENESS

fashionin orderthatby her mannershe mightnot apprisethe examinerof


which were the experimentalbabies.
Certain test procedureswere adopted to increase the reliabilityof the
measurements.We tried as far as possible to keep out of a child's range
of vision until the moment of the test. The examiner recordedonly the
child's firstresponsesand did not prolong the situationbeyond 15 or 20
seconds.Further,the baby was placed upon his back formostsituationsbut
in the prone positionwith his head towardthe sheetforthe "voice" test.
The infants'responsesto each subtestwere recordedon a checkliston
which the possible responseswere set forthunder the categoriesof:
i. Quicknessof regard
2. Duration of regard
3. Kind and durationof facial expression
a. positive:brightening,smiling,laughing; duration
b. negative: soberingor frowning,whimpering,crying; duration
4. Nature of physicalactivity
5. Number of vocalizations.
Scoring. On the basis of preliminaryobservationin the institution
and a studyof the literature,a scoringprocedurefor the social testswas
establishedbeforethe experimentbegan and was used throughoutthe study
withoutchange. Because we were obtainingresponsesto a social stimulus,
a person,we recordedand creditedall responses,thoseof regardand physi-
cal activityas well as those of smiling or vocalizing-the responsesmore
usuallyconsideredsocial. In each categoryitemswere arrangedin an order
which seemed to signifygreaterresponsiveness, fromslownessto rapidity
of response,fromshorterto longerduration,fromless to greateramplitude.
Each item, that is, each step of the scale, was arbitrarilyassigned a value
of one. Because the items were arrangedin ascending order,the score for
any itemwas one plus the value of all otheritemsbelow it in thatcategory.
For example, under the categoryof facial expression,"smiling briefly"
receiveda score of four,since it followed (and was assumed to include)
three items on facial brightening:brighteningbriefly,brighteningrecur-
rently,and brighteningsustainedly.Similarly,the scoresforphysicalactivity
ranged froma one for "moves arms or legs" to a seven for "approach."
The score foreach subtestwas the sum of the scoresforeach category.
The score forthe testwas the sum of the subtests'scores.
A few additionalcommentson the scoringare necessary.Each vocaliza-
tion received one credit up to io, after which no additional credit was
given. A bonus of two points was allowed for vocal tricks,e.g., blowing
bubbles, smacking lips, etc. In the scolding situationonly "negative" re-
sponses were given credit.
The scoringof itemsof negativebehaviorpresenteda difficulty. A neg-
ative responseto a person,enumeratedin the checklistas frowning,whim-

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
EXPERIMENTAL PROCEDURE 17

pering,crying,or physicalwithdrawal,
is a socialresponse;it is not the
absence of a response.Therefore,it did not seem reasonable that it be
scored as zero and certainlynot be assigned a negativevalue. We decided
to give each itemof negativebehaviora value equal to its parallel "positive"
item; thus,whimperingand smilingwere given equal credit.As a result,
the scoreindicatesamountof responsiveness but not its nature;whetherthe
responsewas positiveor negativecould be determined,however,by inspec-
tion of the test blank. Finally, if a child gave both positiveand negative
responses,e.g., smiled then whimpered,he did not receivecreditfor both,
but for only one, whicheverwas assigned the highervalue.
Agreementbetween Observers. To obtain a measure of amount of
agreementbetweentwo different observersof the behaviorsampled by these
tests,the experimenterand the examinermade independent,simultaneous
observationsof the responsesof 13 babies. Table 3 shows the resultsof
these studies.Only absoluteagreementon itemschecked was counted.For
example, if one observerrecorded six vocalizations and the other, eight
vocalizations,the item was counted as a disagreement,even though there
was agreementin part.
Agreementbetween the total scores was also investigated.The rank
ordercoefficient betweentotal scoresbased on itemschecked independently
by the two observersfor the nine subjects on Part A was .88, which is
statistically p being below .oi. For the four subjectsstudiedon
significant,
Part B the total scores for one observerwere
18, 20, 27, and 28; for the
otherobserverthe scoreswere 22, 15, 25, and 29, respectively. Both analyses
show considerableobserverreliabilityin the use of theseinstruments.

TABLE3
PERCENTAGES OF ITEM AGREEMENT FOR THE SOCIAL TESTS

Tests StimulusPerson N Percentage


ofAgreement
Part A ...................... Experimenter.... 4 80, 80, 89, 100
Examiner ....... 5 68, 82, 83, 89, 91
Part B ......... ............. Experimenter.... 2 83, 93
Examiner ....... 2 80, 82

The Other Tests


The test of postural developmentwas composed of items from the
developmentalschedules of Gesell and Amatruda (14), whose procedure
fortestingwas followed.All posturalitemswere set down in
chronological
order as theyappear in the I6-week throughthe 56-weekschedulesunder
the categoriesof supine, prone, pulled-to-sitting,
and standing behavior.
A credit of one was assigned arbitrarilyto each item; half a credit was
allowed for an item the examinerwas not sure of, if it occurredbetween

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
18 MODIFICATION OF SOCIAL RESPONSIVENESS

The scorewas thenumberof itemspassed;highest


successes. possiblescore
was 35. The testwas administered withthe infantin his crib.It always
followed,and neverpreceded, thesocialtests.
The testof cube manipulation was composedsimilarly of itemsfrom
thesamedevelopmental schedules, foronlythecube,notthemassedcube,
situation.The authors'directions were followed.Again,the itemswere
enumerated in chronological order,thistimefromthex2-week through the
56-weekschedule.As before, eachitemwas assigneda valueof one; score
was numberofitemspassed;maximumscorewas 22. This testwas admin-
isteredin a testingroom,and followedthesocialtests.
ScaleforInfants
The CattellIntelligence was givenand scoredaccording
to theauthor'sdirections(8). This scale,too,was givento thebabiesin the
testingroom.
The experimenter was notpresentat anyof thesetests.
RELATIONS BETWEEN TEST DATA AND THE PROPOSITIONS

in termsof
at thispointto restatethehypotheses
It seemsappropriate
thetestdata.
The social testswere designedto measuredirectlythe behaviorwe
wishedto study,namely,the infant'sresponseto an adult. If a child's
scoreincreased,we inferred of the response,
an increasein the strength
and assumed that it representedan increasein social responsiveness.Test
to be measures
scoreswillbe considered oftheamountof socialresponsive-
ness,althoughnot necessarilyof the kind,sincenegativeand positivere-
sponsesreceivedequal values.
Propositionsi and 2 (as statedearlierunderThe Problem)wereac-
cepted on the basis of observationsof the infantsin this institution.These
observationsare reportedunder Results. They constitutethe norm of be-
haviorwhichwe attempted to modify.
The hypotheses to be tested,then,restatedin termsof the empirical
data,are:
I. When theexperimenter presentsherselfas the stimulusperson,the
socialtestscoresof theexperimental subjectsshouldbe greaterthanthose
of thecontrolsubjects.
2. The socialtestscoresof theexperimental subjectsshouldbe greater
in responseto theexperimenter thanto theexaminer or to a third,totally
unknownperson.
3. If (i) and (2) occur,theexperimental subjectsshouldshowsigns
of sobering, frowning, whimpering, crying,or physicalwithdrawal when
theexaminer is thestimulusperson,and evenmoreso whena totallyun-
knownpersonis thestimulusperson.
4. The experimental subjectsshouldshowmoreimprovement in their
scoreson the postural,the cube, and the Cattellteststhan the control
subjects.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS

OBSERVATIONAL DATA

Observationof the subjectsof this study,as well as of otherbabies of


comparable age in this institution,suggested that they awarded special
recognitionto no one person and that none showed the awareness of
strangersdescribed in the literature.There were differencesamong the
babies, to be sure, in their responsivenessand in their interestin people.
But clear, unmistakableevidence of a special kind or degree of response
to any one personwas not seen. Constantquestioningof hospitalpersonnel
and volunteerselicited no examples. With an occasional baby past io
monthsof age who was a favoriteof a memberof the hospitalpersonnel,
the experimentersaw such evidence, in a steadinessof gaze, an alertness
of facial expression,vocalizations directedto the person, and a reaching
out of arms. The most remarkablecharacteristic of most of the babies was
their sociability.Babies were observed to smile at people who were not
looking at them.They smiled to people in the corridorseen throughwin-
dows. In nine monthsof almost daily observationwe saw only one baby
who did not smile to people.
If these infantswere aware of the strangenessof any person who ap-
proached them,again, the experimentersaw no clear, unmistakablesigns.
Some babies had periods of fussiness;some, of course,cried. Some volun-
teersgot along betterwith some babies than others.But none of the babies'
reactionsseemed to arise fromtheirreactingto an adult's being a stranger
to them. They smiled to visitorsin streetgarb, althoughthose who cared
for them always wore standard hospital gowns. They appeared to be as
friendlywith the occasional male visitorsas with the more frequentfemale
ones.
At this point it seems appropriateto reportcertainobservationsabout
the experimentalbabies. Although,as we shall see, a differencein their
responsivenessto the experimenterappeared in their test scores afterone
week of treatment,the experimenterherselfwas not sure until the fifth
or sixthweek that theygreetedher arrivaleach morningwith any
special
recognition,the kind of smilingeagernessa baby shows his mother.At the
beginningthe experimenterwas concernedthat the babies might become
so attached to her that they would adjust poorly to the hospital routine
aftershe lefteach day and over the weekend.The hospitalhad instructions
to call her at any timeif theythoughtshe was needed. Actually,thesefears
were unnecessary.The general opinion of the hospitalpersonnelwas that
the experimentalbabies were more placid and better-natured than the
control babies and made fewer demands. At the end of the treatment
period two experimentalbabies in each experimentwere reportedto have
been difficult to handle,and one of thesein the firstexperiment
apparently
19

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
20 MODIFICATION OF SOCIAL RESPONSIVENESS

wasveryupset,
but,interestingly
enough, didnotappearuntil
thedistress
Monday, the thirdday afterthe experimenterleftand the day she might
havebeenexpected
back.
Needless to say, the experimenterbecame deeply attached to the ex-
perimentalbabies.
thehealth
Finally, ofbothcontrol
andexperimental wasgood
subjects
throughoutthe periodof studywith the exceptionof colds. All the subjects
had excellentappetitesand in general this was characteristic
of the babies
in the institution.

COMMENTSON TREATMENT
OF THE DATA

I. It will be recalled that in each experimentthe eight babies were


divided into fourpairs roughlymatchedon the basis of chronologicalage,
IQ, and initial social responsiveness.But subjectswere removed fromthe
institutionand as each left his match was lost, if the analysis was to be
limitedto a studyof change withinpairs. At the eighthweek threepairs
from the firstexperimentwere still intact,but only one pair from the
second experiment.At the end of the twelfthweek only one pair from
each experimentwas intact. We decided, therefore,to work with group
means.
2. Throughoutthe followinganalysisof the data the resultsof the two
parallel experimentshave been pooled. No evidence has been discovered
to suggestthatthesubjectsof thetwo experiments were not samplesfromthe
same population.Table 4 showstheconsiderableresemblancebetweenthetwo

TABLE4
MEANS AND RANGES OF VARIABLES USED FOR MATCHING
IN THE TWO SEPARATE EXPERIMENTS
(N equals 8 foreach experiment)

Experiment1 Experiment2
Item Mean Range Mean Range

Chronological
age ........... 6.1 5.7-6.7 6.2 5.1-7.4
Cattell IQ ................. 98.9 75-119 92.1 75-110
Socialscore................ 27.6 18- 34 28.0 12- 43

samples in means for chronologicalage and some pretestscores,although


there is greatervariabilityamong the subjects of the second experiment.
The experimentaltreatment,in all importantdimensions,was the same
for the two groups. Time-samplingstudiesdemonstratedsubstantialagree-
ment (see Table 2). More meaningfulevidencethatthese probablyare not

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 21

heterogeneoussamples may be found in Figure 7 which shows veryclose


agreementbetweenthe responsesof the experimentalsubjectsof the parallel
experimentsto the experimenterover the course of treatment.
3. In thefirstexperimentthepretestresponsesof onlysome of the babies
to the examinerwere obtained. Missing are the scores for four babies on
Part A of the social test,as well as the scoresforall eightbabies on Part B.
In onlythe analysisof variancestudies,whereinitialmeasureswere desired,
we substitutedfor the missingdata the scoresmade by the subjectsin re-
sponse to the experimenter. The reasoningwas thatat the pretesting, before
the experimentalconditionof any subjectwas determined,and, of course,
beforetreatmenthad begun, there should be considerableagreementbe-
tween the responsesof a baby to two, as yet,neutralpersons.In the second
experimentpretestscoreswere obtained for all babies in responseto both
experimenterand examiner.

NOTE ON PRESENTATIONOF THE DATA

In the followingsectionwe will present,first,a summaryof the data


togetherwith curves based upon all the scores available at each testing
period over the 12 weeks of the study.Statementsdescribingthese curves
are tentative,of course,and are presentedas speculationsfor which verifi-
cation might be sought in experimentsspecificallydesigned to test them.
In generalthesespeculationsare apart fromthe hypothesesthisexperiment
was designedto test.
Following this presentationthere appears, where appropriate,a statis-
tical test of the significanceof the results.These tests do bear upon the
propositionsset forthin the introduction.Most of them are based not upon
the totalsample but upon a smallersample of six experimentaland six con-
trol subjects,the greatestnumber for whom there were scores over the
period of experimentaltreatment.The means of these 12 cases are dis-
played in the tables togetherwith the means for the total sample; the
differences betweenthe two samplesappear to be small. The testmost fre-
quentlyemployedis the analysisof variance.This techniqueseemed justi-
fied because thereappeared to be no marked departurein the data from
the assumptionsupon which the testis based. The model used throughout
the studyis Lindquist's typeI (22).

THE SOCIAL TEST

Responseto Experimenter and Examiner


Table 5 presentsthe results obtained on the social test as a whole.
Scores are sums of Parts A and B (7 subtests).It will be recalledthat the
parts were given on alternateweeks, except for the pretests.Scores then

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
22 MODIFICATIONOF SOCIAL RESPONSIVENESS
are sumsofscoresofPartA at week I and PartB at week 2, of PartA at week
3 and Part B at week 4, etc. Only subjectsforwhom thereare scoresat both
weeks have been included.

TABLE 5
MEANS AND RANGES OF SOCIAL TEST SCORES

Experimental
Group ControlGroup
Periods N Mean Range Mean,* N Mean Range Meana

A. Experimenter, the
StimulusPerson
Pretests ...... 8 56.8 25- 97 59.7 8 55.4 49- 65 54.5
Weeks 1 and 2 8 74.1 64- 84 75.5 8 52.0 44- 71 53.8
Weeks 3 and 4 6 74.8 65- 88 .... 8 54.6 38- 67 53.2
Weeks 5 and 6 6 73.5 60- 88 .... 7 61.1 48- 82 57.7
Weeks 7 and 8 6 74.0 62- 91 .... 6 51.2 39- 64
Weeks 9 and 10 5 86.8 60-105 .... 5 59.8 51- 66
Weeks11 and 12 5 83.2 54-112 .... 3 66.0 57- 74
B. Examiner, the
StimulusPerson
Pretests ...... 4 62.2 26- 87 57.8t 4 56.0 52- 60 52.2t
Weeks 1 and 2 8 55.1 39- 67 55.7 8 50.5 39- 65 53.5
Weeks 3 and 4 6 62.8 50- 75 .... 8 59.6 49- 70 58.3
Weeks 5 and 6 6 63.5 52- 81 .... 7 56.9 49- 77 58.2
Weeks 7 and 8 6 65.7 57- 84 .... 6 57.7 45- 68
Weeks 9 and 10 5 70.8 62- 86 ... 5 53.0 38- 64
Weeks11 and 12 5 72.2 64- 80 .... 3 69.0 58- 81 ....

*Meana is mean of the six casesused foranalysisof variance.


t See text.

Figure I shows that the curves for the experimentalgroup in response


to both stimuluspersons,the experimenterand the examiner,exceed those
of the controlgroup. Further,the curve for the experimentalgroup in re-
sponse to the experimenterincreasesfairlyrapidlyin the firsttwo weeks,
maintainsits heightduring the remainingsix weeks of experimentaltreat-
ment,thenrisesagain aftertheend of thetreatment. The curvefortheexperi-
mentalgroupin responseto the examiner,on the otherhand, risesgradually
afteran earlydecrease.One may speculatethatat firstan increasein respon-
sivenessto one personis associatedwith,perhapsproduces,a decreasein re-
sponsivenessto anotherperson,but that,with time,increasedresponsiveness
to one personis associatedwith increasedresponsiveness to another,though
smallerin amount. The curves for the controlgroup in responseto both
personscrossat pointsand until the last testingperiod are fairlystable.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 23

80

"wo o
I/-I

50
--- E's TO EXPERIMENTER
S---. E's TO EXAMINER
o---o GC' TO EXPERIMENTER
-.-o C'sTOEXAMINER
PRETEST land2 3and4 5and6 Tand8 9andlO Ilandl2
TESTING PERIODS IN WEEKS
FIGURE 1-Means of Social Test forexperimental
and controlgroupsin response
to experimenterand examiner.

It should be notedthatnumberof cases decreasedwith timeand thatthe


range of scoresat each testingperiod was wide, with overlappingof scores
betweenexperimentaland controlgroups.
An analysis of variance of scores for the smaller sample (six experi-
mental and six controlsubjectsat fivetestingperiods,as describedabove)
is presentedin Table 6. The analysissuggeststhat it is veryunlikelythat
the differencebetween the experimentaland the control subjects in re-
sponse to the experimentercould have arisen by chance, although the
differencein responseto the examinermay have. "Periods" had no statis-
ticallysignificanteffectupon the scores,nor did the interactionof treat-
ment with periods. It appears thereforethat forthese subjectsit was treat-
ment and not the passage of time which produced a statisticallysignificant
effect.
Althoughby this testthereappears to be no statisticallysignificantdif-
ferencein the responsesof experimentaland controlsubjectsto the exam-
iner, the curvesfortheseresponsessuggestthe presenceof some difference.
A t test of paired comparisonsbased on the means for all experimental
and controlsubjectsat each testingperiod yields a value of 5.11, with p

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
24 MODIFICATION OF SOCIAL RESPONSIVENESS

TABLE 6
ANALYSIS OF VARIANCE OF SOCIAL TEST SCORES FOR
EXPERIMENTAL AND CONTROL GROUPS

as Stimulus
Experimenter Examineras Stimulus
SourceofVariation df MS F p MS F p

Betweensubjects .... 11
Treatment ......... 1 4684.4 38.12 <.001 395.2 3.40 >.05
Error (b) .......... 10 122.9 116.4

Withinsubjects ..... 48
Periods ............ 4 167.6 1.00 >.05 142.2 1.02 >.05
Treatment X periods 4 132.1 0.79 >.05 13.3 0.09 >.05
Error(w) .......... 40 167.7 139.3

Total ............. 59

less than .005. (Differencesbetweenmeans were adjusted forN of cases by


weighting.) We interpretthese findingsto mean that there is some ten-
dency for the experimentalbabies to become more responsiveto the exam-
iner.
Finally, we can testfor the effectof stimulusperson by a t testof the
differences betweenthe means for experimentaland controlgroups using
for each subject the sum of the five measures. The experimentalbabies
were more responsiveto the experimenterthan to the examiner (t equals
8.96, p is less than .ooi); the control subjects were no more responsive
to the one than to the other(t equals 1.o07,
p is more than .I).

Responseto the Stranger


At the eighthweek, at the end of treatment, a person unknown to the
babies servedas stimulusobjects in both experiments.Subtestsi and 2 of
Part A only were given,and thesescoreswere comparedwith similarscores
made in responseto the experimenter and to the examiner.Table 7 shows
means for responsesof the experimentaland controlgroups to the three
persons.Figure 2 gives the actual distributionof scores,fromwhich it can
be seen thatthe lowestscoremade by an experimentalsubjectto the experi-
menter,i.e., 31, is above the means forall othergroups.
These data suggestthat the experimentalsubjectswere more responsive
to the experimenter than to eitherof the othertwo stimuluspersons; that
the experimentalsubjectswere more responsiveto all three persons than
the controlsubjects;and that the experimentalsubjectswere as responsive
to a "stranger"as to the examinerwith whom theyhad experience.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 25

TABLE 7
MEAN SOCIAL SCORES OF EXPERIMENTAL AND CONTROL
S GROUPS TO THREE PERSONS AT WEEK 8
(N equals 6 in each group)

StimulusPerson Group
Experimental ControlGroup

Experimenter.......... 37.2 24.2


Examiner ............. 28.2 25.0
Stranger .............. 28.0 23.2

50
0 * EXPERIMENTAL
o CONTROL
40 X MEANSCORE
x 0

Xw x _0 00
L o

20 o00

cO

w w

wI (D
0: w
: z z o
0. 4 - x.
W Lu ) w w co
STIMULUS PERSONS
FIGURE 2-Responses to three stimulus persons at Week 8.

Table 8 gives the resultof an analysisof varianceof these scores.The


increasein the social responsivenessof the experimentalsubjectsto all three
persons appears to be statisticallysignificant,although the differencesin
their responsivenessto the three stimulus persons might have arisen by
chance. The differences in the responsivenessof the experimentalsubjects

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
26 MODIFICATIONOF SOCIAL RESPONSIVENESS
and the examiner,
to the experimenter and to the experimenter and the
stranger, appearnotto be attributable
however, to chance,t testsbetween
thesepairsof meansyieldingprobabilities
of lessthan5 percent.

TABLE 8
ANALYSIS OF VARIANCE OF SOCIAL SCORES OF EXPERIMENTAL AND
CONTROL SUBJECTS TO THREE PERSONS AT WEEK 8

Sourceof Variation df MS F p

Betweensubjects............ 11
Treatment................. 1 441.0 7.18 <.05
Error (b) ................. 10 61.4
Withinsubjects............. 24
Stimulus .................. 2 87.0 1.91 > .05
TreatmentX stimulus...... 2 83.1 1.82 >.05
Error (w) ................. 20 45.6
Total ..................... 35

TABLE9
MEANS AND RANGES OF PART A SCORES FOR EXPERIMENTAL
AND CONTROL GROUPS

Experimental
Group ControlGroup
Periods N Mean Range Meana* N Mean Range Meana

A. Experimenter,the
StimulusPerson
Pretest ..... 8 31.2 12 - 48 33.2 8 32.8 26 - 41 31.5
Week 1 .... 8 41.4 25-53 41.3 8 31.9 16-47 32.0
Week 3 .... 8 47.0 37- 59 46.0 8 32.4 22 - 37 31.5
Week 5 .... 7 44.0 36 - 49 44.2 8 35.6 30- 50 32.5
Week 7 ... 7 50.3 38 - 63 49.0 6 33.0 23 - 49
Week 9 .... 6 44.3 35 - 54 .... 5 35.6 29 - 43
Week11 .... 5 50.2 32-62 .... 3 38.7 31-43
B. Examiner, the
StimulusPerson
Pretest ..... 7 33.0 12 - 56 34.8 5 29.2 19 - 36 28.7
Week 1 .... 8 34.5 18 - 49 32.8 8 26.8 11 - 37 26.0
Week 3 .... 8 36.1 27 - 51 36.8 8 32.4 23- 42 32.3
Week 5 ... 7 37.3 26-57 39.2 8 31.4 17-43 32.8
Week 7 .... 7 37.1 23 - 55 39.5 6 31.3 16 - 44
Week 9 .... 6 41.2 30-51 .... 5 32.2 24-44
Week11 .... 5 45.8 37 - 55 .... 3 36.0 33 - 38
"*Mean, is meanof the six casesused foranalysisof variance.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 27

Performanceon Part A
For purposes of analysis the performanceof the subjectson each part
of the test will be consideredseparately.Table 9 gives number of cases,
means, and ranges of Part A scores for the responsesof the experimental
and controlgroups to the experimenter and to the examinerat each testing
period. The curves are shown in Figures 3 and 4, and resemblesomewhat
the curvesforthe testas a whole. We see thatas earlyas the firstweek the
experimentalsubjectsexhibitedan increasein social responsivenessto the
experimenter;in fact,the largestgain occurredin this period. The slow
rise in scoresforthe experimentalsubjects'responseto the examineris also
apparent.

60

50

I . ,
I 4

--- EXPERIMENTAL
GROUP

10 - CONTROLGROUP

L L.J L. .J L. .J ..
PRETEST I 3 5 7 9 II
TESTING PERIODS IN WEEKS
FIGURE
3--Meansand rangesof Social Test, Part A, wvhen
the experimenter
is
thestimulusperson.

An analysis of variance of scores for the smaller sample (described


above) is presentedin Table Io. The analysissuggeststhat it is very un-
likely that the differencesbetween the experimentaland control subjects
in responseto the experimenter could have arisen by chance, althoughthe
differencesin responseto the examinermay have. "Periods" had no statis-
effectupon the scores,nor did the interactionof treatment
ticallysignificant
with periods.However, a t testof paired comparisons(as describedforthe

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
28 MODIFICATION OF SOCIAL RESPONSIVENESS

60

50

40 -"

PRETEST 1 3 5 7 9
TESTING PERIODS IN WEEKS
FIGURE 4-Means and rangesof Social Test,Part A, when the examineris the
stimulusperson.

whole test) suggeststhat here too thereis some tendencyfor the experi-
mental babies to become more responsiveto the examinerthan the control
subjectsdid, when we testover the full courseof the experiment(t equals
8.64,p is less than .0005).

TABLE IO
ANALYSIS OF VARIANCE OF SOCIAL TEST, PART A, SCORES FOR
EXPERIMENTAL AND CONTROL GROUPS

as Stimulus
Experimenter Examineras Stimulus
Sourceof Variation df MS F p MS F p
Betweensubjects .... 11
Treatment ......... 1 1696.0 28.70 <.001 614.4 4.07 >.05
Error (b) .......... 10 59.1 150.9
Withinsubjects ..... 48
Periods ............ 4 125.6 1.59 >.05 92.4 1.02 >.05
Treatment x periods 4 95.2 1.20 >.05 5.2 .06 >.05
Error(w) .......... 40 79.0 90.2
Total ............. 59

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 29
We testedalso forthe effectof stimuluspersonby using foreach subject
the sum of his scores frompretestthroughweek 7 (as was done for the
analysis of the test as a whole). The experimentalbabies were more re-
sponsiveto the experimenterthan to the examiner(t equals 3.89, p is less
than .oi); the controlsubjectswere no more responsiveto the one than to
the other (t equals 0.72, p is greaterthan .2).

Performanceon Part B
Table i gives the numberof cases, means,and rangesof Part B scores
for both groups to the stimuluspersonsat each testingperiod. These data
are shown in Figures 5 and 6. As with Part A scores,the experimental
subjectsare more responsiveto the experimenter than are the controlsub-
jects, but the differencebetween the curves appears to be smaller. The
curves show a less consistentpatternof performance.The experimental
subjectsgain over the firsttwo weeks, and again betweenthe eighth and
tenth weeks, that is, after the end of treatment.The responsesof both
groups to the examinerresembleeach other closelyover the course of the
experiment.Again, the range at many testingperiods is wide.

TABLE II
MEANS AND RANGES OF PART B SCORES FOR EXPERIMENTAL
AND CONTROL GROUPS

Experimental
Group ControlGroup
Periods N Mean Range Meana* N Mean Range Meana

A. Experimenter,the
StimulusPerson
Pretest ..... 8 25.5 12 - 49 26.5 8 22.6 17 - 26 23.0
Week 2 8 32.8 25 - 44 34.2 8 20.1 12 - 27 21.8
Week 4 6 28.8 20 - 43 .... 8 22.2 12 - 34 21.7
Week 6 6 29.3 11 - 40 .... 7 26.1 16 - 37 25.2
Week 8 .... 6 25.0 15 - 31 ... 6 18.2 12 - 26
Week10.... 5 40.6 23 - 55 .... 5 24.2 19 - 32
Week12 .... 5 33.0 22 - 50 .... 3 27.3 26 - 29
B. Examiner, the
StimulusPerson
Pretest ..... 4 22.8 14 - 31 23.0 4 24.0 20 - 28 23.5
Week 2 .... 8 24.2 16-31 22.8 8 23.8 10-41 27.5
Week 4 .... 6 26.0 20 - 36 .... 8 27.2 20 - 40 26.0
Week 6 .... 6 24.3 10- 36 .... 7 27.7 14-45 25.3
Week 8 .... 6 26.2 20- 34 .... 6 26.3 23- 33
Week10 .... 5 27.6 22 - 38 .... 5 20.8 14 - 28
Weekl2 .... 5 26.4 23 - 34 .... 3 33.0 25 - 44
* Meana is mean of the six casesused foranalysisof variance.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
30 MODIFICATION OF SOCIAL RESPONSIVENESS
60

50

40

! i

10 EXPERIMENTAL
GROUP
"CONTROL
GROUP

..L-J Lj L.J L L..J .


PRETEST 2 4 6 8 10 12
TESTING PERIODS IN WEEKS
FIGURE 5-Means and rangesof Social Test, Part B, when the experimenter
is
the stimulusperson.

50

40

20 b-O--

"EXPERIMENTAL
GROUP
CONTROL GROUP

PRETEST 2 4 6 8 10 12
TESTING PERIODS IN WEEKS
FIGURE6-Meansand ranges
of SocialTest,PartB, whentheexaminer
is the
stimulus
person.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 31
Table 12 showsan analysisof varianceof thesix experimental and six
controlsubjectsforwhomtherewerepretestscoresas well as foursubse-
quentscores,coveringtheeightweeksof treatment. Althoughp now lies
between.05 and .oi fortheeffectof treatment upon the experimentalsub-
jects'responses to theexperimenter
(insteadof below .ooi as forPart A),
it is stillunlikely
thatthedifferences
betweentheexperimental and control
subjectscouldhave arisenby chance.No othereffects are statistically
sig-
nificant.

TABLE 12
ANALYSIS OF VARIANCE OF SOCIAL TEST, PART B, SCORES FOR
EXPERIMENTAL AND CONTROL GROUPS

as Stimulus
Experimenter Examineras Stimulus
Sourceof Variation di MS F p MS F p

Betweensubjects.... 11
Treatment ......... 1 693.6 15.05 <.05 24.1 0.60 >.05
Error (b) .......... 10 46.0 40.1
Withinsubjects ..... 48
Periods ............ 4 75.6 1.25 >.05 16.9 0.37 >.05
Treatment X periods 4 36.4 0.60 >.05 11.3 0.24 >.05
Error(w) .......... 40 60.6 46.1
Total ............. 59

Testingfortheeffectof stimuluspersonby usingforeach subjectthe


sumofhis scoresfrompretest through week8 we findthaton thissecond
partof the as
test, on thetestas a wholeand on PartA, theexperimental
subjectsweremoreresponsive to the experimenter
thanto the examiner
(t equals 3.15,p is less than .025), while the controlsubjectsdid not respond
differently to eitherperson (t equals 0.54, p is more than .2).
In summary, thereappearsto be considerable differencein the scores
yieldedby the two partsof the social tests.It will be recalledthatno
effort
wasmadetoequatetheparts.The explanation maylie in thedifferent
kindsof responseselicitedby the two parts.But it is possiblealso that,
to someextent,onlythe circumstancesof administering thesubtests were
forthe difference.
responsible The problemis considered in moredetail
under Performanceon Subtests.

Performanceon the Parallel Experiments


It seemedappropriate,
becauseof thesmallsize of the sample,to dis-
covertheextentof agreement
betweenresultsobtainedon thetwoparallel

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
32 MODIFICATION OF SOCIAL RESPONSIVENESS

experiments,performedone afterthe other. If thereappeared to be some


resemblance,the confidenceto be placed in the findingswould be increased.
Figure 7 shows the curves for the two groups of experimentalsubjects
on Part A in responseto the experimenter.(N is only four and becomes
progressivelysmaller at succeeding testing points). Although other sets
of curvesdid not show as close a resemblance,nowherewas theremarked
disagreement.The amount of agreementis the more noteworthywhen we
recall that the second experimentwas more rigorousboth in assignment
of subjectsto treatmentconditionand in concealingfromthe examinerthe
experimentalstatusof the subjects.
We conclude thereforethat the relativelysimilar performanceof the
two groups reinforcesconclusionsbased on the pooled sample.

60

50

(I)o
40

30 d

EXPERIMENT I
20m ------- EXPERIMENT 2

I I I I I I
PRETEST I 3 5 7 9 II
TESTING PERIODS IN WEEKS
FIGURE groupson Social Test,Part A, in responseto
7-Means of experimental
experimenter.

Performanceon Subtests
Each of the seven subtestsappeared to make a differentcontributionto
the total score. Subtestsi and 2, while not yieldingidenticalcurves,both
showed a responseto treatmentin the early weeks. Subtests3 and 4 re-
sponded to treatmentonly moderatelyat first;the largestimprovementset
in in the weeks followingthe end of treatment.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 33
Subtestsof Part B appear not to discriminatebetweenthe experimental
and controlbabies as well as subtestsof Part A. From experiencein admin-
isteringsubtests2 and 3 of Part B we suspectthat of all the social situa-
tions these were most affectedby the procedureof having two persons
presentwhen the experimenter was the stimulusobjectand onlyone person
presentwhen the examinerwas the stimulusobject.When the experimenter
held a baby in her arms,the baby tendedfirstto look at the examinerwho,
if she was to observethe baby's responses,was a prominentobject,no mat-
ter how she triedto stayin the background.In this subtestthe advantage
lay with the examinerwho had no competitorforattentionwhen she held
the baby in her arms. And in subtest3 the experimenterstared straight
ahead while the examinerobservedthe child's behavior;the examinerhad
to peep at the baby when she presentedthe situation,and fromexperience
we noted that the babies became more active as soon as they caught a
person'seye.

"Negative" Responses
Frowningor worriedlook, whimpering,crying,or physicalwithdrawal,
including turns head away, rolls to side away from person, or crawling
away, were definedas negativeresponses.It should be recalledthatnegative
responsesreceivedscores equal in value to the so-calledpositiveresponses,
smiling,laughing,and physicalapproach. A negativeresponseto an adult
is a definitesocial response;it shows, presumably,that the adult possesses
meaning for the child. The positive-to-negative
dimensionof social respon-
sivenessmay be thoughtof as cuttingperpendicularly the dimensionless-to-
more social responsiveness.
To only 65 of the 527 subtestsadministeredin the whole study was
a negative response recorded. Responses to the "scolding" situation of
Part A are not included here. Of the 65, 26 were incidentsof rolling of
the head or body to the side away from the adult. These often seemed
to us not attemptsto withdrawfrom,or to shut out, the stimulusperson,
especially since most of them occurred following smiles or laughs, and
usually the baby rolled back promptly.No child crawled away; in only
threesubtestsdid a baby frown,leaving only 36 subtestsin which a child
whimperedor cried,a less debatable item of negativebehavior.And only
27 of these 36 were cryingor whimperingwhich was not preceded or
followedby smiles.The conclusionis thatnegativebehaviorwas infrequent,
accountingforonly 5 to 12 per cent of the responses,dependingupon how
inclusiveis the definitionof negative.
Negative responsesare part of the problembeing investigated,that is,
the relationshipof recognitionof the motherto an awarenessof
strangers.
It was predictedthat if the experimentalsubjectsbecame more
responsive

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
34 MODIFICATIONOF SOCIAL RESPONSIVENESS
to the experimentertheywould in timeshow an awarenessof strangers
in negativefacial,vocal,or bodilyresponses.
Table 13 showsnegativeresponses givenby experimental and control
childrento theexperimenter and to theexaminer. The totalnumberhere
is 56 becausethecountnow is of thenumberof timesany subjectgave
a negativeresponseoverthewholecourseof testinginsteadof numberof
situationsin whicha negativeresponsewas given.On Part A the two
groups performed similarlyto thetwopersons.On PartB, however, twice
as manyexperimental childrengave negativeresponses, but again they
wereequallydividedbetweenthe two stimuluspersons.Since thesere-
sponsesrepresentso smalla fraction of the totalresponses,
theyprobably
do notwarrant moreintensive Certainly
analysis. thehypothesisis notsup-
ported.

TABLE 13

NEGATIVE RESPONSES OF EXPERIMENTAL AND CONTROL GROUPS


TO EXPERIMENTER AND TO EXAMINER

Part A Part B
StimulusPerson E Group C Group E Group C Group Total

Experimenter.......... 7 9 7 3 26
Examiner ............. 9 9 8 4 30
Total ........... . 16 18 15 7 56

Threechildrenaccountedfor25, or almosthalf,of theresponses. Two


ofthesewereexperimental babies.
No controlchildgave a negativeresponse to thestranger,but threeof
theeightexperimental childrendid: one criedrecurrently, two criedsus-
These resultscould supportthe hypothesis
tainedly. thatincreasedattach-
mentto one personproducesan awarenessof thestranger, exceptthatthe
threewereall membersof thefirstexperiment; no experimental babyin
the secondexperiment a
gave negativeresponse to the This
stranger. dis-
crepancybetweenthe two experimental groupsmakesus reluctant to say
thedata supportthehypothesis.

Vocalizations
It will be recalledthatthenumberof vocalizationswererecordedfor
each subtestand a value of one assignedto each up to a totalof Io in
each subtest.Figure8 givescurvesformeannumberof vocalizations on
all subtests forexperimentaland controlgroupsin responseto theexperi-
menterand the examinerseparately.It shouldbe notedthat,to obtain

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 35
numberof vocalizationsforall subtests,vocalizationsof Part A were com-
bined with thoseof Part B given the followingweek. Figure 8 shows that
the numberof vocalizationsgiven by the experimentalgroup to the experi-
menterincreasedover the 12 weeks of the experiment,that the numberof
vocalizationsgiven by this group to the examiner decreased during the
eight weeks of experimentaltreatmentbut increased slightlyduring the
four weeks followingthe end of treatment.The curves for the control
group in responseto both the experimenterand the examinerare similar.
Afteran increasein the middle of the experimenttheyfall offsharplyand
continueto decreaseso that at the end of the 12 weeks thereappears to be
considerabledifferencebetween number of vocalizationsgiven by the ex-
perimentalgroup to eitherperson and number of vocalizationsgiven by
the controlgroup to eitherperson.Experimentaltreatmentthen seems to
have produced an increase in the number of vocalizationsutteredby the
experimentalbabies during the social tests.These findingsare related to
thoseof Brodbeckand Irwin (4) who showed thatinfantsrearedin institu-
tionsvocalized less than familybabies.

AE GROUP
TO
.,A IEXPERIMENTER

o \ E GROUPTO EXAMINER
/ ...,

SC TOEXAMINER%
GROUP
z C GROUPTO EXPERIMENTER...
I I I I I I "
PRETEST I and2 3ond4 5ond6 7and8 9ondIO
TESTINGPERIODS IN WEEKS Iand12

FIGURE 8-Mean numberof vocalizationsof experimental


and controlgroups
in responseto experimenter
and examiner.

No statisticalanalysisof theseresultswas performedbecause no


hypoth-
esis had been set up originallyconcerningthe effectof treatment
upon
vocalization. The results suggest,however, that number of vocalizations
might providean index to the effectof treatment.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
36 MODIFICATION OF SOCIAL RESPONSIVENESS

TEST OF POSTURALDEVELOPMENT

Table 14 shows the means and ranges of scoreson the posturaltestfor


all subjects,while the curvesforthe means of the experimentaland control
groups are shown in Figure 9. The most prominentcharacteristic of these
resultsis the increasewith age. Experimentaland controlgroups,although
not matched on posturalachievement,performedat about the same level
on the pretests.Thereafterthe means for the experimentalgroup exceed
those of the control group at every testingperiod but one-that at the
third week, where the control group mean is seven-tenthsof one point
higher.
TABLE14
SCORES ON THE POSTURAL TEST FOR THE EXPERIMENTAL AND CONTROL
GROUPS AT ALL TESTING PERIODS

Experimental
Group ControlGroup
Periods N Mean Range Meana* N Mean Range Mean,
Pretest....... 8 16.4 9 -30 17.8 8 16.1 10 -23 15.5
Week ...... 8 17.1 11 -30 18.6 8 15.4 10 -22.5 14.8
Week 3 ...... 8 18.7 12.5-32 20.4 8 19.4 12 -26 18.9
Week 5 ...... 7 25.2 20 -34 25.1 8 21.8 14.5-27.5 22.1
Week 7 ...... 7 28.1 21 -35 27.7 6 24.9 15 -30
Week 9 ...... 6 28.8 24 -30 .... 5 24.8 16 -30
WeekII ...... 5 30.2 27 -30 .... 3 28.0 24 -30
"*
Meanais meanofthesixcasesusedforanalysis
ofvariance.

An analysis of variance of scores made by the six experimentaland


controlsubjectsfor whom there are scores at each period over the eight
weeks of treatmentsuggeststhat the slightlybetterperformanceof the
experimentalsubjectsis not statistically
significant.The differenceamong
the periods,however,is significant,p being less than .ooi.
Sittingand Standing
Because many of the items in the supine and prone situationsof the
posturaltest were passed by a majorityof the babies and thereforewere
not discriminative,
and because the experimentalbabies had more oppor-
to
tunity practice sittingand standing (they were more often seated in
chairs and more often held standing in the experimenter'slap), scores
on the sittingand standing items of the test were analyzed separately.
Figure io shows thatperformanceon theseitemsdoes not differmarkedly
fromperformanceon the test as a whole. The experimentalgroup made
considerableprogressin standingbetweenthe thirdand fifthweeks, while

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 37

30

25

w
o A/
"0 20

1,5 EXPERIMENTAL GROUP


" , CONTROL GROUP
15-

PRETEST I 3 5 7 9 II
TESTING PERIODS IN WEEKS
FIGURE9-Means of testof posturaldevelopment.

EXPERIMENTAL
-CONTROL

I00
CA SITTING 4

STANDING do

PRETEST I 3 5 7 9 II
TESTING PERIODS IN WEEKS
FIGURE 10-Means of sittingand standingitems.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
38 MODIFICATIONOF SOCIAL RESPONSIVENESS
thecontrolgroupmadeverylittle,butthecontrolgroupshoweda similar
thoughsmallerspurtin the nextperiod,betweenthe fifthand seventh
weeks.

TESTOFCUBEMANIPULATION
Table 15 givesthe meansand rangesof scoreson the testof cube
behaviorforall experimentaland controlsubjects.
The meansincreasewith
time,morerapidlyat the beginningof the experiment and withsome
levelingoffafterthesixthweek(whichmaybe a function ofthetestitself).
The experimental groupturnedin a betterperformance on thistestat
the initialtestingperiod,and overthe courseof testingmaintained just
aboutthesameamountofsuperiority. Thereis no evidencefrominspection
of the curvesforcube behavior(Figure Ix) thatthe experimental treat-
mentmodified theinitialdifference
betweenthegroups.

TABLE 15
SCORES ON THE CUBE TEST FOR THE EXPERIMENTAL AND CONTROL
GROUPS AT ALL TESTING PERIODS

Experimental
Group ControlGroup
Periods N Mean Range Meana* N Mean Range Meana

Pretest ....... 7 8.3 5-12 8.8 8 5.8 1- 11 6.0


Week 2 ...... 8 9.0 3-14 10.3 8 8.4 4-12 9.0
Week 4 ...... 6 13.2 8-18 .... 8 10.8 4-14 10.7
Week 6 ...... 6 15.2 12- 19 .... 7 13.0 7-16 12.8
Week 8 ...... 6 14.5 12- 17 .... 6 12.0 6-15
Week10 ...... 5 16.6 15- 19 .... 5 14.2 13-15 ....
Week12 ...... 5 14.0 11-16 .... 3 12.3 14-15 ....
0
Meana is meanof the six casesused foranalysisof variance.

Analysisof varianceof cubescoresforthesix experimentaland control


subjectsovertheeightweeksconfirms theinformation alreadypresented.
Treatment seemsto havehadlittleeffect,
p ofF ratiobeinggreater
than.o5,
butthepassageof timeappearsto accountformuchof thevariancein the
scores,since"periods"as a maineffectyieldedan F ratioof 32-78,p less
than.oor.

THE CATTELL INFANT INTELLIGENCESCALE

The CattellInfantIntelligence
Scale was givenjust beforethe experi-
mentbeganand,onceagain,duringtheseventhweekof treatment. Means
and rangesof IQ are shownin Table i6 for thosechildrenwho were

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
THE RESULTS 39

15

TESTING PERIODS IN
TESTING PERIODS IN WEEKS
WEEKS
FIGURE 11-Meansof testofcubemanipulation.

stillin the institution


at the seventhweek of the experiment. The data
suggest thattreatmentincreased the ofthe
slightly performance experimental
group,whileitsabsenceresulted in a slightdecreaseforthecontrolgroup.
But thedifference betweenpretestand finalIQs is not statistically
signifi-
cantforeithergroup,t being.15 fortheexperimental groupand .62 for
thecontrolgroup.
The rangeof IQs decreasedon retestforbothgroups.This decrease
maybe simplya regression ofscoresto themeanor mayreflect thedecreas-

TABLE 16

MEANS AND RANGES OF CATTELL IQ'S FOR THE EXPERIMENTAL


AND CONTROL GROUPS
(N equals 6 in each group)

Experimental
Group ControlGroup
Periods Mean Range Mean Range
Pretest .................... 97.5 75-119 94.5 75-108
Week 7 ................... 98.3 92-103 91.3 85- 97

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
40 MODIFICATION OF SOCIAL RESPONSIVENESS

ing variabilityin scores shown by Bayley (3) to be characteristicof the


period fromsix to 12 monthsin a comparativestudyof severalscales and
with different samplesof infants,but not includingthe Cattellscale.
It is interesting
to note,incidentally,
thatthe average IQs of both groups
fall withinthe normalrange.We excluded,of course,infantswith marked
physical defect-actually one blind and one spastic child-and excluded
one baby on the basis of mentaldefect,one who had an IQ of 50. There is
not sufficientevidenceto declare thatthe IQs of the untreatedchildrenare
falling,but certainlythroughthe eighthmonth these subjectsas a group
do not resemblein IQ the institutionpopulationsreportedin the literature.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
DISCUSSION

In this experimentthe chiefelementof the experimentaltreatmentwas


the substitutionof one "mother" for many "mothers."The primarygoal
of the treatmentwas to bring about an increase in social responsiveness
to this one personand to measurethe effectsof this learningupon respon-
siveness to other persons. But in substitutingone "mother" for many
"mothers" other dimensions of the environmentwere changed as well.
The one "mother"was with the childrenmore of the time; she did more
for them; and, coming to know the babies, she was more readilyable to
take into account theirpreferences, and general reac-
dislikes,irritabilities,
tion patterns.The resultsobtainedin thisstudy,therefore, cannotbe related
only to the substitutionof one "mother"for many,or to any othersingle
componentof the treatment.
It would be possible,now that an increasein social responsivenesshas
been demonstrated to thiscomplexindependentvariable,to assess the effects
of varyingcomponentsone by one. The amount of time spent with the
children by one or many "mothers" could be held constant, or the
multiple"mothers"could increasethe amount of time spentwith the chil-
drenuntilit equalled thatspentby the one "mother."Similarly,the number
of caretakingacts could be matchedacrossthe two groups.The one "moth-
er" might be persuaded to care for the babies in fixed order as hospital
routinedictated;or, perhapsmore satisfyingly forthe caretakersand more
rewardingly forthe babies, the many "mothers" could be instructedto care
forthe babies in accordancewith theirapparentneeds.
The design of this experiment,further,does not cast any light on the
mechanisms,that is, the reinforcers, responsiblefor the learning.The in-
fantsmay have become more responsiveto the experimenterbecause she
acquired secondaryrewardvalue as the gratifierof theirneeds (9, 15, 27).
Within the categoryof needs,it may be thatthe satisfactionof some needs
was more importantthan the satisfactionof others.Or, the babies' respon-
siveness may have been brought about by the experimenter'srewarding
their social responses-by paying attention,by smiling and vocalizing in
return,by playfullyand affectionately handling them. It was this kind of
behavioron the part of the infantto which she would be most responsive.
On the other hand, Dennis (9) has shown that infants,during the first
26 weeks of life,smile to adults even when the smile is not respondedto.
Our impression,based upon day-by-day observationof the infantsduring
the experiment,is that the experimenterwas an effectivestimulus,not so
much in termsof the servicesshe was performingfor the child, that is,
whetherfeeding,bathing,or changing diapers,etc., but in relationto the
attentionshe gave and receivedfrom the child. The attentionseemed to
bk not just a general awareness as might well occur
during a caretaking
41

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
42 MODIFICATION OF SOCIAL RESPONSIVENESS

sequence,but a directand clear act of attending.In any one day therewere


manysuch activeinterchangesof attendingbetweenchild and experimenter,
sometimesduring, but often apart from,caretakingactivities.Generally,
these interchangeswere playful or affectionatein nature. Could this se-
quence of responsesbe analyzed and made the subjectof experimentapart
fromthe caretakingactivities,we expectit would yield some answersto the
question of the mechanismsresponsiblefor social learning.
The experimentaltreatmentoriginallywas thoughtof as an enriching
of the environment,particularlyas it might reduce maternaldeprivation
(16). The psychologicaldeprivationof an institutionhas been portrayed
often.But as the experimentprogressed,it became less certainthat what
the experimenterprovided could be unequivocally characterizedas en-
richmentof all areas of experience.For, as the experimenterwas accustom-
ing the experimentalbabies to her presenceand mannerof care, the control
babies, on the otherhand, were seeing many people and having to adjust
to being fed, bathed,and handled in a great varietyof differentways. We
suggest only that as the experimentaltreatmentprovided certain experi-
ences, it also excluded or reduced otherexperiences.
Furthermore,the experimentaltreatment,although it seemed both
extensiveand intensiveduring the process of carryingit out, probably
constitutedonly a small segmentof an already complex life, when con-
sidered against all that had occurredin the infant'slife beforetreatment
began, all that occurredafterthe experimenterleft each day, and on the
weekends.
We turn now to the chief question this experimentwas designed to
answer: the effectof increased responsivenessto a known person upon
responsesto otherpersons.It was predictedthat as the infant'sdiscrimina-
tion between known and unknown persons developed he would show
signs of discomfortwhen approached by unknown persons. But in this
studysuch signs did not accompanythe discrimination. If "by naturesome
infantsare more sensitiveto strangersthan others" (15, p. 262; 17), it
seems unlikelythat all eight experimentalbabies would belong to the less
sensitivegroup. More likely,theirpast experiencewith many people, and
theirdaily experiencewith some other people, are responsiblefor the ab-
sence of the behaviorcalled awarenessof the stranger.
Not only did the experimentalbabies not show signs of discomfort
when approachedby strangepersonsbut insteadtheybecame more friendly,
accordingto some of the evidence. One might say simplythat a learned
response has been transferredfrom the experimenterto other similar
stimuli.An explanationwhich takes into account processas well as results
can be considered:that the experimentaltreatmentset up a continuously
reinforcingenvironment.If it is true,as hospital personnelreported,that
the experimentalbabies were more placid and better-natured, they them-

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
DISCUSSION 43
selvesmayhave favorablyalteredtheresponsestheyreceivedfromthewomen
who cared forthemwhen the experimenter was not in the institution.
Finally, the resultsof this study appear to confirmthe hypothesisthat
social behaviorin the infantis modifiableby environmentalevents.In con-
trast,postural,"adaptive," and "intellectual"behavior proved to be less
modifiableby the experimentaltreatment.It is true that no special efforts
were made to train the babies in these areas or even to give them special
opportunityto practice.But neitherdid the experimentaltreatmentoffer
systematictraining in social responsiveness.Certainly the experimental
babies were out of theirbeds more often,were given toysmore frequently,
etc. On the otherhand, with more done for them,the experimentalbabies
may have had less need to do forthemselves.But if this were a significant
variable,the controlbabies,who were more oftenleftto themselves,should
have surpassed the experimentalbabies. The experimentalenvironment
may have provided both more opportunityto practice and less need to
practice,and the interactionof these variables may have been responsible
for the lack of a statistically
significantdifference
in theirperformances.It
seems reasonableto conclude that the institutionprovidedan environment
as adequate forthe developmentof thesebehaviorsas did the experimental
treatment.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
SUMMARY

The purpose of this studywas to explorethe effectupon infantsof an


experimentalmodificationof the environment.For a group of institutional
babies one "mother" was substitutedfor many "mothers"and the effect
upon social responsiveness was assessed.This was the primaryfocusof the
study, but the effectsupon other kinds of behavior were also measured.
The main question asked was, "What effectwill an increasein social re-
sponsivenessto one person,the 'mother,'have upon responsiveness to other
persons?"
The subjectswere i6 infantsabout six monthsof age, who lived in an
institutionwhere they were cared for by constantlychanging "mothers."
Eight of these infantsformedthe experimentalgroup. They were cared
for,four at a time, by the experimenteralone during a daily period of
seven and one-halfhours,fivedays a week, for eight weeks, a total of 300
hours.The experimenter triedin so far as possibleto adapt her care to the
individual needs of the experimentalsubjects.A controlgroup, each time
numberingfourinfants,was cared forby the usual hospitalroutine.There
were thus two parallel experimentsperformedone afteranother.
Systematictime-samplingof the activitiesof caretakersin the experi-
mental and the controlrooms made by an observerthroughoutthe eight
weeks of each experimentaltreatmentdemonstratedthat the experimenter
by herselfperformedmany more caretaking acts for the experimental
subjectsthan a large numberof different women performedforthe control
subjects.For example,in the firstexperimentcaretakingacts were recorded
for 23 per cent of the observationson the experimentalroom, all of them
performedby the experimenter, while in the controlroom caretakingacts
totaled only 7 per cent of the observations,and these acts were divided
between 14 different persons.Further,the controlsubjectswere more often
withoutany adult in theirroom and spentmore time in theircribs.
Both experimentaland control subjects were given a batteryof tests
beforethe experimentbegan and at weekly intervalsfor the eight weeks
of the experimentaltreatmentand also for the four followingweeks. To
eliminateany biases the experimenter mightacquire, all testswere admin-
isteredby anotherperson,the examiner,whom the experimentertried to
keep in ignoranceof the dimensionsof the experimentalconditionand of
the hypothesesto be tested.The testsused were a testof social responsive-
ness, tests of postural developmentand of cube manipulation,and the
Cattell Infant IntelligenceScale.
The social testwas composedof seven subtests,fourin Part A and three
in Part B. The two parts were not equated, the second part being used
to yield extended observations.Each subtestwas designed to obtain the
infant'sresponsesto such natural situationsas an adult's approach to his
44

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
SUMMARY 45

crib, holding him in arms, and returninghim to his crib. The infant's
responseswere obtained to three stimulus persons: to the experimenter,
a known person to the experimentalsubjects; to the examiner,a relatively
unknown person to all subjects (the infantssaw her only weekly as the
tester);and to a completelyunknownpersonwho approachedall the babies
in only the eighth week of the experiment.All of the infant'sresponses,
includingquicknessof regard,durationof regard,facial expressions,physi-
cal activity,and vocalizations,were recordedand scored by a point scale.
Scores were obtainedforsubtests,and summedfora totalscore."Negative"
behaviors,e.g., frowning,crying,physicalwithdrawal,were assignedvalues
equal to parallel "positive" behaviors (smiling, laughing, physical ap-
proach).
An increase in a child's score was consideredto representan increase
in the strengthof the responseand, hence, to indicatean increasein social
responsiveness.The nature of the response,whetherpositiveor negative,
was determinedby inspectionof the testblank. Independentsimultaneous
observationsmade by the experimenter and the examinerin a preliminary
study showed considerableobserverreliabilityfor the test.
The posturaland cube testswere composed of normativeitems taken
fromthe Gesell DevelopmentalSchedules,arrangedin chronologicalorder.
It was predictedthat the experimentalsubjects would become more
sociallyresponsiveto the experimenterthan the controlsubjects; that the
experimentalsubjects would become more responsiveto the experimenter
than to the examineror to a person totallystrange;that the experimental
subjects,if theybecame more responsiveto the experimenter, would show
an awarenessand probablyfear of the strangenessof the examinerand of
the third,unknown,person. It was predictedalso that the experimental
subjectswould show more improvementin theirscoreson the postural,the
cube, and the Cattelltests.
The resultsindicated that the experimentalsubjectsdid become more
socially responsiveto the experimenterthan the control subjects (p less
than .ooi); that the experimentalsubjectsbecame more responsiveto the
experimenter than to the examiner(p being less than .oi); thatthe experi-
mentalsubjectsappeared to becomemore,ratherthan less, responsiveto the
examiner, although the differenceis not always statisticallysignificant;
finally,therewas no evidencethat the experimentalsubjectsbecame afraid
of the examiner.Similar resultsappeared in responseto a totallyunknown
personat the end of the eighthweek.
Considerable agreementbetween the resultsof the social testsfor the
two parallel experimentsservesto increasethe confidencewith which these
conclusionsmay be accepted.
Responses of all babies were predominatelypositive, "negative" re-
sponses occurringin less than 12 per cent of all responses,about equally

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
46 MODIFICATIONOF SOCIAL RESPONSIVENESS
dividedbetweenexperimenter and examiner. The numberof vocalizations
heardduringthe socialtestsincreasedoverthe courseof the experiment
for the experimental subjects,especiallyin theirresponseto the experi-
menter.For thecontrolsubjectsthe numberof vocalizations appearedto
decreaseafterthefifth weekin responseto boththeexperimenter and the
examiner.
The experimental subjectsmade slightly higherscoreson thepostural,
cube,and Cattelltestsbutthedifferences werenotstatistically
significant.
The chiefelementof theexperimental treatmentin thisstudywas the
substitutionof one "mother" formany"mothers." But in theprocessother
dimensions of theenvironment werealso changed.Our impression, based
uponday-by-day observationof theinfants,is thatthe mechanism respon-
sibleforthelearning was notthecaretaking perse,butratherthefrequent,
active,and usuallyplayfulinterchanges of attentionbetweenexperimenter
and baby whichoccurredduring,but oftenapart from,the caretaking
activities.
That theincreasein socialresponsiveness to one personshownby the
experimental babies did not produce an awareness of the strangenessof
others,as was originally predicted, may be attributedto theirpast and
recurrentexperience withotherpeople.Theirincreased friendliness
toothers
suggestsa transferof learningor a processin whichtheythemselves favor-
ablyalteredthe responses theyreceivedfromothercaretakers duringthe
courseof theexperiment.
Finally,theresults appearto demonstrate thatthesocialbehaviorof the
infantmay be modified byenvironmental events.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
S

REFERENCES

1. BAYLEY, NANCY. A studyof the cryingof infantsduring mentaland


physicaltests. J.genet.Psychol.,1932, 40, 306-329.
2. BAYLEY, NANCY. The California First-Year MentalScale. Univer. of
SyllabusSeries,1933,No. 243-
California
3. BAYLEY, NANCY. Consistencyand variabilityin the growth of intel-
ligence frombirthto eighteenyears. J. genet. Psychol.,1949, 75,
165-196.
4. BRODBECK,A. J.,& IKwIN, O. C. The speechbehaviorof infantswith-
out families. Child Develpm., 1946, 17, 145-156.
5. Bi3HLER,CHARLOTTE.The firstyear of life (trans. Greenberg,Pearl,
& Ripin, Rowena) New York: Day, 1930.
6. BifHLER, CHARLOTTE. From birthto maturity. London: Kegan Paul,
1935.
7. BiUHLER,CHARLOTTE,& HETZER, HILDEGARDE. Testing children'sde-
velopmentfrombirthto school age (trans. Beaumont,H.) New
York: Farrar& Rinehart,
1935.
8. CATTELL, PSYCHE. The measurementof intelligenceof infantsand
young children. New York: Psychol.Corporation,1940.
9. DENNIS,W. An experimentaltestof two theoriesof social smilingin
infants. J. soc. Psychol.,1935, 6, 214-223.
o0. DENNIS, W. Infantdevelopment underconditionsof restricted
prac-
ticeand ofminimumsocialstimulation.Genet,Psychol.Monogr.,
1941,23, 143-I89.
i. DENNIS,W., & DENNIS,MARSENAG. Behavioral developmentin the
firstyear of life as shown by fortybiographies. Psychol. Rec.,
1937, I, 349-361.
12. I)ENNIS, W., & DENNIS, MARsENA G. Development undercontrolled
environmental conditions. In W. Dennis (Ed.), Readings in child
psychology. New York: Prentice-Hall,1951.
13. FREUD, ANNA, & BURLINGHAM, DOROTHY. Infants without families.
New York: Int. Univer.Press,1944.
14. GESELL, A., & AMATRUDA,CATHERINE S. Developmental diagnosis.
New York: Hoeber, 1941.
15. GESELL,A., & THOMPSON,HELEN. Infant behavior,its genesis and
growth. New York: McGraw-Hill, 1934-
16. GOLDFARB,W. Effectsof psychologicaldeprivationin infancyand
subsequentstimulation. Amer. J. Psychiat.,1945, 102, 18-33.
47

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions
48 MODIFICATION OF SOCIAL RESPONSIVENESS

17. GOODENOUGH, FLORENCE. Developmental psychology. New York:


Appleton-Century, 1934-
18. GRIFFITHS,RUTH. The abilitiesof babies. New York: McGraw-Hill,
1954-
19. HEBB,D. O. On the natureof fear. Psychol.Rev., 1946, 53, 259-276.
20. HEBB, D. O., & RIESEN,A. H. The genesis of irrationalfears. Bull.
Canad. Psychol.Ass., 1943, 3, 49-50.
21. JERSILD, A. T. Child psychology.(4th Ed.) New York: Prentice-
Hall, 1954-
22. LINDQUIST, E. F. Design and analysis of experimentsin psychology
and education. Boston: Houghton-Mifflin, 1953-
23. MALRIEU, P. La constructionde l'objet et les attitudessociales de
l'enfantde la naissance'i deux ans. i. Psychol.norm.path., 1951,
44, 425-437.
24. PIAGET, J. The constructionof realityin the child (trans. Cook, Mar-
garet) New York: Basic Books, 1954-
25. SHIRLEY,MARYM. The firsttwo years,a studyof twenty-five babies.
Vol. II. Intellectualdevelopment. Minneapolis:Univer. of Minne-
sota Press, 1933-
26. SPITZ, R. A. Hospitalism. An inquiryinto the genesis of psychiatric
conditionsin earlychildhood. Psychoanal.Study of Child., 1945,
", 53-74-
27. SPITZ, R. A. The smilingresponse:a contributionto the ontogenesis
of social relations. Genet. Psychol.Monogr., 1946, 34, 57-125-
28. STONE,L. J. A critiqueof studiesof infantisolation. Child Develpm.,
1954, 25, 9-20.

This content downloaded from 139.86.7.217 on Sat, 17 Oct 2015 10:46:50 UTC
All use subject to JSTOR Terms and Conditions

You might also like