Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Influence of a Consistent Minority on the Responses of a Majority in a Color Perception Task

Author(s): S. Moscovici, E. Lage, M. Naffrechoux


Source: Sociometry, Vol. 32, No. 4 (Dec., 1969), pp. 365-380
Published by: American Sociological Association
Stable URL: http://www.jstor.org/stable/2786541 .
Accessed: 05/04/2011 03:07
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=asa. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
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.

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to
Sociometry.

http://www.jstor.org

Influenceof a ConsistentMinorityon the


Responses of a Majority in a Color
PerceptionTask
S. MOSCOVICI,* E. LAGE AND M. NAFFRECHOUX
Ecole Pratique des Hautes Etudes, Paris
social
Most of the studieson social influencehave dealt with conformity,
pressureexercisedby majoritygroups,and have used dependencyas the
source of influence.This study concernsinnovation,social pressureexercized by a minority,
and triesat the same time to prove that behavioral
is
a
style
generalsource of influence.An objectivelyblue stimulusis used
whichtwo subjects (stooges) out of six call "green" in the experimental
groups.When the behaviorof the minorityis consistent,the numberof
"green" repliesin the experimental
groupsis significantly
higherthan in
the controlgroup.This changein answeris not onlya verbalagreementbut
corresponds
to a changein theirperceptioncode, as shownby a color discrimination
test. Whenthe minority'sbehavioris not consistent,
its impact
on the majorityis minimal.Thereforeit is the consistentbehavioralstyle
of minorities
thatinsuresthe adoptionof theirpointof view.
THE CONFORMITY BIAS

assimilatesthe processof influenceto the


Specialisedliteraturecommonly
processof conformity
(Allen, 1965). On the one hand, the tendencyis to
assume that any type of influenceleads to conformity,
and moreoverthat
is the sole phenomenon
conformity
achievedby means of influence.On the
otherhand,whenexaminingtheindividual,it is alwaysassumedthathe asks
himselfthequestion"ShouldI followthegroupor theminority?"
or in other
wordshe is faced with the alternativeof conformity
or deviance.On the
an individualfrequently
contrary,
poses the questionin exactlythe inverse
manner:"WhatshouldI do so thatthemajoritywilladoptmypointof view?
of such posHow can I changethe conceptionof others?"The multiplicity
sible questionstendsto contradictthe aforementionedassimilation.Without
goinginto the detailsstatedelsewhere(Moscoviciand Faucheux,1969) we
can considerthe innovationas a formof social influence.
In orderto study
thisform,theanalysisof theactionof a minority
and empirically
theoretically
upon the majority,the qualitieswhichit mustpossessin orderto make its
*Fellow (1968-1969) at the Center for Advanced Study in the Behavioral Sciences.
I also wish to acknowledgethe assistance extendedto me by the James Marshall Fund.
365

366

SOCIOMETRY

This research
point of view accepted,constitutesa sort of prolegomenon.
proposesto show more clearlyone of these qualities and to depart from
the customaryemphasison attitudeswhichare linkedto conformity.
BEHAVIOR STYLE AS A SOURCE OF INFLUENCE

In almostall of the researchdone to date on social influenceonly one of


deand experimentally:
its possiblesourceshas been studiedtheoretically
pendency.
Nonetheless,forcertainreasons,we cannotmake use of it in the study
of innovation.Firstof all, it seemsclear that dependencyin relationto an
individualor a subgroupwhichinnovates,is a consequenceratherthan a
The necessityto heed the
cause of an actionaimedat exertingan influence.
computeror televisionexpertsfollowsthe adoptionof
advice of electronics,
or television,or any kind of specifictechelectronicequipment,computers,
social
nical invention.A minoritywhichtrulyinnovates,whichtransforms
reality,only rarelyhas powerat the outset.In addition,it is to be noted
are
thattheindividualsor subgroupswho changerules,values,or knowledge,
not judged as being superiorto othersinsofaras competenceis concerned.
In short,dependencyin relationto the phenomenonwhich interestsus
factorwhichcan
variable,nor a differential
is neithera decisiveindependent
account for influencewhich is exerted.Thus, we were promptedto seek
whichis not subjectto the limitations
whichwe
anothersourceof influence
and whichcomescloserto expressingtheactiveresolute
have just mentioned,
We believethatwe have foundit in the behavioral
characterof a minority.
style of the individualor those individualswho propose a solutionto a
problem,a new normfora group.Good reasonsexistto supposethatin the
process of innovation,the way in which the behavioris organizedand
to provokethe acceptanceor the rejectionof a judgpresentedcould suffice
Moreover
mentor a proposedmodelduringthe courseof social interaction.
the consistencyof the behaviorof a minority,the fact that it resolutely
maintainsa well definedpointof viewand developsit in a coherentmanner,
whichunderthe
appearsas if it oughtto be a powerfulsourceof influence,
wouldnot be a resultof an explicitdependency.
circumstances
made by one of the authorsin collaborationwith
A seriesof experiments
and
Moscovici,
1967) has alreadyshownthe impactof a con(Faucheux
sistentminorityupon a majoritywhen preferencejudgmentsconcerning
of an implicitnormare involved.
equiprobablestimulior the modification
of thepreviousone, we should
a
is
which
continuation
In thepresentstudy,
the majoritynormto be
when
is
also
this
action
that
possible
like to prove
changedis explicitor quasi-physical.
Why are we expectingsuch an effect?The presenceof a normcan be dis-

INFLUENCE OF A CONSISTENT MINORITY

367

in the spontaneousunanimityof thosewho share it, and in the


tinguished
expectancythat a high probabilityresponsewill occur in the face of a
stimulusor a determinate
object. The validityof judgmentsand opinions
(Kelley, 1967) and the stabilityof relationswiththe environment
are guaranteedowingto thisnormonlyif thesetwo criteriaare expected.
Now, let us supposethat a subgroupdivergesfromthis customarymode
of responseand that he providesan alternativemode of responseto the
in
same object,the same stimulus.The diversitywhichreplacesuniformity
and of conflict;doubt is cast upon
the groupis a creatorof uncertainty
of responsesof each personor of the groupand the variability
thehierarchy
is increased.By insistingon his answer,a minoritywill not only engender
a conflict,
the conflict,
because it poses its own judgments
but will intensify
and opinionsas havingthe same value, as beingequivalentto thoseof the
majority(Worell 1967). Moreover,this insistenceproves that takingone's
stand is not casual and that the subgrouphas no intentionof concedingor
submittingto the group.
This exertsa tremendouspressuretowardsacceptanceof the new and
surprising
response. We mustalso add that theseconflictrelationsassume
a particularcharacterin thecase wherethe stimulusis physical.The reality
to be judged in thesecircumstances
is not individual,arbitrary:it is common,in principleuniversal.No matterwho, facedwithsuch a reality,one
is expectedto react in the same way, and each one imaginesthat he is
reactingas he is supposedto react.
In an experiment
cited by Asch (1962), Sperlingdemonstrated
that the
influenceexertedon an individualis muchgreaterwhenhe believesin the
existenceof an objective response,than when he does not believe in it.
Thus, the fact that a physical stimulusis involveddoes not necessarily
workagainst the exertionof influenceby a minority;on the contraryit
may facilitateit. The majorityhas one singlemeans to reducethe tension,
to ignorethe judgmentof the minority:that is to transform
the conflict
This means that it mustbe able
of responseinto a conflictof attribution.
not as being producedby the propertiesof the
to explain the difference
stimulus,but as being producedby those who perceiveit: an anomalyof
vision,a lesser judgmentcapacity. This is possible when minorityis an
isolated individual(Moscovici 1969).
In the eventthatnothingin the situationpermitssuch an attribution
and
a dyad, cannotbe distinguished
that membersof the minority,
constituting
frommembersof the majorityby such traits,thenthe latterare even more
obligatedeitherto adopt the responseof the minorityor to reject it, i.e.,
to polarize.No othermeans is left to themto restorethe invariability
of
responsein theirrelationwiththe externalworld.

368

SOCIOMETRY

the influence
in mind,in orderto demonstrate
With thesepresuppositions
of a minorityupon a majoritywithina group, we have conceivedan
experimentin which:
and
of the minority
(a) Responseconflictis increasedby the consistency
by the consensusamongits members.
(b) Objectivityis an implicitexigencyof judgments.
are exclusive,constituting
(c) The responsesof themajorityand minority
withouteitherone just negatingthe other,as, forexample,
an alternative,
if one wereto say that two unequal amountsof dots weresaid to be equal.
in judgmentcannot be accountedfor by individual
(d) The difference
to be composedof more
qualities. (Thus it was necessaryforthe minority
than one person.) Otherwisethe conflictin responsecould be transformed
to be explained by
into a conflictof attribution,permittingdifferences
forexample.
personaleccentricities,
(e) The judgmentof the majorityin the laboratoryis identicalwiththat
of any randomsampleoutsidethe laboratory,so that the judgmentof the
can be expectedto be directlycounterto the normalexpectations
minority
in society.
EXPERIMENTAL PROCEDURE AND RESULTS

FIRSTEXPERIMENT. The subjectswereliberalarts,law and social science


materialfemalesubjectswere
students.Giventhenatureof the experimental
in evaluatingthe colorof an
because of theirgreaterinvolvement
preferred
types of
object. The stimuliused consistedof slides with two different
the passage of a beam
filtersmountedin them: (1) photofilterspermitting
of light of the dominantwave length (X=483.5) in the blue scale;
(2) neutralfilterswhichreducedlightintensityin certainproportion.
In a set of six slides,threeslides were moreluminousthan threeothers.
These variationsin lightintensitywere studiedin orderto make the task
was controlled.
morerealisticand less boring.Theireffectin thisexperiment
Each experimental
group consistedof fournaive subjectsand two confederates.Once the subjects were seated in a row before the screen on
whichwere to be projectedthe slides, theywere told that this would be
an experimenton color perception.At the same time they were informed
that theywouldbe asked to judge the colorand variationin lightintensity
of a seriesof slides (a briefexplanationof the meaningof lightintensity
was furnished).Beforepassinga judgment,the whole groupwas asked to
in orderto checktheparticipants'"chromatic
takea Polack testcollectively,
sense."
This test had a twofoldobjective: first,to eliminatethose subjectswho

INFLUENCE OF A CONSISTENT MINORITY

369

perchancemighthave visual abnormalities;second,to emphasizethe fact


that everyonein the group had normalvision,so that the confederates'
in vision,i.e., to a personal
responsewill not be attributedto a difference
factorexternalto the experimentalsituation.
Afterthe collectivecorrectionof the resultto the test,and afterhaving
ascertainedthat everyonesees normally,the subjectswere instructedwhat
responsesmightbe given and how the experimentwould be conducted,
to wit replyingaloud and naminga simple color as well as estimating
the lightintensityin numericalterms(rangingfrom0 forthe dimmestto
5 for the brightest).Subjects were also told that the preliminarytrial
would be just forpracticein whicheach subject would only make a light
intensityjudgment.
The real purposeof thesepreliminary
trialswas to enable the subjects
to get acquaintedwiththe color of the stimulusand to immunizethemin
McGuire's (1964) sense of word against the futureonslaughtof the instructedminority
whichdoes not share the norm.During thesepreliminary
trials the confederatesansweredat random. Following these trials, the
series of six different
slides was presentedsix times, the order of the
slides varyingsystematically
fromone seriesto the next. Thus these were
36 trials,each one lasting15 seconds,separatedby approximately
5 seconds
of darkness.In each trialthe two confederates
exertedinfluenceby calling
the color "green." In this manner,the confederateswere both internally
consistentfromone trial to the next with each other,since they gave all
the timethe same response.
At the end of the experiment
the subject filledout a questionnaireconcerningthe stimuliand the othermembersof the group.As usual, the real
were explainedbeforeleaving the room.
objectivesof the experiment
Two variationswere introducedregardingthe seating of the two confederatesand the presentation
of the stimuli.
variation:in 12 groupsthe confederates
(1) Confederate
wereseated side
by side and gave the firstand second responses,while in the 20 other
groupstheywere separated,and occupiedthe firstand fourthplaces. The
variationin the seatingof the second confederate
was aimed at modifying
of his behavior,that is to say, to make him appear
the interpretation
more independentof the firstconfederate.
(2) The stimulusvariation:in orderto test the impactof the commitment to the firstresponseand to permita possible change,we modified
the mode of presentationof stimuli.In 13 groupswhich included those
were seated in position1 and 4, the continuity
in whichthe confederates
of the sequence of the stimuliwas interrupted
by introducingtwo oneminutepauses aftera sequenceof 12 slides.

370

SOCIOMETRY

The orderof responseof the subjectsremainedthe same fromone trial


to the next forthe durationof the experiment.
We wonderedwhetherthe subjectsexperiencedan
SECONDEXPERIMENT.
influencewhich,even if it did not resultin a change in verbal response
did have a lasting effecton theirperception.We
duringthe experiment,
expecteda shiftin the blue-greendesignationthresholdwhichwould reveal
a reactionthatwas repressedduringthe social interaction.Certainsubjects
did refuseto adoptopenlytheminority
response,feelingcompelledto remain
loyal to the generalnorm,even when they themselvesbegan to doubt its
itselfby an
validity.Here one mightexpecta latentattractionmanifesting
extensionof the designation"green" to stimuliin a zone whicha control
groupwould call blue. The oppositereaction(extensionof the notionblue
to stimuliin the greenzone) would be the resultof polarization.
is identicalto the precedingexperiment,
The firststageof thisexperiment
exertsits influenceon the majority.At the
thatis to say that the minority
thanked subjects telling them that
end of this phase the experimenter
anotherresearcherwho was also interestedin visionphenomena,would like
of the
to solicittheirparticipationin anotherresearchproject,independent
one in whichtheyhad just participated.He left the roomand the second
enteredimmediately
and repeatedhis request.The latterhavexperimenter
ing obtainedthe agreementof the subjectsseated themarounda table and
relatedto the effectof the exercise
said to themthat it was an experiment
about the visionphenomena.He then describedthe material,isolated the
subjectsby means of cardboardscreensand instructsthemto writedown
the responsesindividuallyon a sheet of paper. The materialconsistedof
16 disks in the blue-greenzone of Farnsworth100-hueset perceptiontest.
Three disks fromeach end of the "blue" and "green"scale wereabsolutely
unambiguous,
but theother10 stimulimightappearambiguous.Afterhaving
made sure that the subjects understoodthe instructions
well, the experimenterannouncedthe beginningof the test. Each disk was presentedon
a neutralbackgroundfora periodlastingapproximately
5 seconds; it was
placed in the centerof the table so that it would be visible to everyone.
The seriesof 16 disks was presented10 timesin the continuousmethod.
The order of presentationwas randomized.Afterthe discrimination
test
the firstexperimenter
returned,the subjectsfilledin the postexperimental
and the experiment
endedin the same manneras the previous
questionnaire
one.
Ten groups participatedin this experiment.
whichwas identicalto the first
THaRD EXPERIMENT. In this experiment
the consistencydegreeof the confederates.
In this
one, only we diversified
case theyanswered24 times"green" and 12 times"blue," the dispersion

INFLUENCE OF A CONSISTENT MINORITY

371

of "blue" answersbeing randomized.Eleven groups participatedto this


experiment.

For thisgroup
The controlgroupwas thesame forthe threeexperiments.
the presentation
of the stimuluswas continuous.The controlsubjectsalso
phase.
took,of course,the discrimination
test afterthe initialexperimental
In all we had 22 controlsubjects,or fourgroupsof 6 subjects,with the
response
eliminationof two subjectswho failed to give the discrimination
accordingto the instructions.
RESULTS
"Green" responses(responseswhichexpressthe
influenceof minority
in the experimental
groups) constituted8.42 per cent
of the answersof the 128 naive subjects in the two firstexperiments.
There is no significant
difference
betweenthe two series of groupson the
questionnaire.Amongthe 22
perceptiontestsnor on the postexperimental
two
greenresponses,representsubjectsof the controlgroup,only one gave
subjects.That means
ing 0.25 per centof the responsesof the uninfluenced
that the latterperceivedthe stimulusas reallyblue and that this normis
firmlyestablishedsocially.
The difference
betweencontroland experimentalsubjects on the basis
of Mann Whitney'sU test (Z=2.10) turnsout to be significant(p=.019,
one-tailedtest). Other data show this influenceas well. Subjects changed
theirresponse (giving 4 or more green responses) in 43.75 per cent of
the groups.The percentageof individualswho yielded was 32 per cent.
Thus we have two categoriesof groups,those in whichno subjectswere
influencedand those in which subjects were influenced.In the latter,it
can be seen that 57 per cent of the subjectsor two subjectsper groupon
the average gave the same responseas the confederates.18.70 per cent
greenresponseswere obtainedin these groups.
Thus, the quantityof green responseswhich we obtained was not so
much the resultof isolated individualswho followedthe confederate,as
of judgmentwithinthe group.The confederates'
theresultof a modification
or discontinuous
seatingposition,and the typeof introduction-continuous,
effect.
-of the stimulidid not have any differentiation
Moreover,we have noticed that even thoughno color contrasteffect
when light inexisted,the subjectswere more similarto the confederates
were
weak
than
when
were
tensities
strong(Z=3.37, p<.003, Mannthey
concernU
This
with
the
Bezold-Briicke
phenomenon
Whitney test).
agrees
different
luminosities.
with
of
color
of the
Yet, irrespective
ing perception
was
of
significantly
higherin the
luminositythe proportion greenresponse
in
the
control
than
experimental
groups
groups.
THE PERCEPTUAL TASK.

SOCIOMETRY

372

In thethirdexperiment,
whereone or severalresponsesof theconfederates
we obtainedonly 1.25 per cent greenresponses.A similar
wereinconsistent,
proposalwas obtainedin groupscompletely
inconsistent
(50 per centblue50 per cent green responsesof the confederates).Althoughwe have to
inconexploremoresystematically
the variationof inter-and intra-subject
sistency,the resultswe have just mentionedare suggestiveof a marked
influenceof the behaviorstyleof a minority.
THE DISCRIMINATION TEST. The question here concernswhetherthe
subjects who changed their social response under the influenceof the
consistentminority
also changedtheirperceptivecode. In addition,we also
wanted to verifythe hypothesisthat the subjects who did not change
theirsocial response,even in the groupwherethe majoritywas not at all
influencedat this level by the minority,
at least changedtheirperceptual
code.
The measurementof the thresholdmakes it possible to verify this
hypothesis.Our calculationsbear on the thresholdvalues, whichwere obtained by a graphicmethodon the smoothedout curve of individualresponses.We retainedthreevalues: (1) the 50 per cent thresholdindicating
the point in the orderedsequence of stimuliwhere the subject gives as
many"blue" as "green"judgments;(2) the lowerthresholdvalue indicates
the point wherethe subject gives 75 per cent greenand 25 per cent blue
judgments;and (3) the upper thresholdvalue, where the subject gives
25 per cent greenand 75 per cent blue judgments.To study the influence
of the consistentminority,
we subsequentlyeliminatedthe resultsof three
subjects in the experimentalgroups who polarized. Their 50 per cent
thresholdwas lower than that of all the controlgroup thresholds.It was
theirlowerthresholdvalue, whichindicatesa generalizationof the notion
of blue in the greenzone. Then,by comparingthe 50 per cent,75 per cent,
and 25 per cent thresholdsof the experimental
groups (37 subjects) and
the controlgroups(22 subjects) we obtained(Table 1) the expectedshift.
All of the data reflectthe effectof interactionbetweenminorityand
of theperceptualcode. This modification
affects
majorityin themodification
TABLE 1
Shiftin the ThresholdforPerceptionof the Color Green
Threshold
50
75
25

ControlGroup
SD
Mean
47.39
46.16
48.41

1.21
1.42
1.14

ExperimentalGroup
Mean
SD
48.03
46.85
49.19

1.38
1.54
1.28

t
1.78
1.68
2.33

P (one-tailed
level)
.038
.047
.01

INFLUENCE

OF A CONSISTENT

MINORITY

373

more subjects than the change of verbal responses.This propositionis


supportedby other data. On the one hand, if within the experimental
groupsa distinctionis apparentbetweensubjects who sometimesadopted
the minorityresponseand subjects who never adopted the minorityresponses,no such difference
emergesin the discrimination
test for the three
thresholdsunderconsideration.
On the contrary,it must be observedthat
shiftis even morepronouncedforgroupswherethe majoritydid not change
than it is forthosewhereit changed,and the Student'st of 1.50 is close
to the 1.68 value,whileit would be significant
at .10.1
We had made the assumptionthat in the groupswhere therewas no
changein social response,or wherethe "green"responsehad been in some
way "repressed"one would observea greaternumberof "green"judgments
in the discrimination
test. One can see that this is indeed the case. The
difference
between the groups where the majoritydid not change and
wherethe majoritydid changeis significant
(X2= 14.94, p<.002). We can
concludethat the consistentminorityhas an even greaterinfluenceon the
perceptivecode of the subjectsthan on theirverbalresponseto the slides.
Of coursethe experimental
techniqueemployedwas not withoutits faults.2
But the resultsobtained should be mentionedonly for the new research
line it gives us.
THE POSTEXPERIMENTAL
QUESTIONNAIRES.
The postexperimental
quesThe
tionnaires
we had devisedshowedus that: (a)
divergence
of opinionor
responseof the consistentminorityconstrainsthe subjects to a cognitive
activitybearingupon the stimulus.The perceptivechangeis not produced
by a pure attractiontowardsthe minority.(b) The relativecertaintyof
the majorityis probablyweakenedas a resultof the confrontation
with
the minority,and its problemwas to explain not why it followedthe
minority,
but why it did not followit.
(a) The CognitiveActivityof The ExperimentalGroup. To begin with
we can put forwardthat occasionallyseeinggreen slides, or seeing green
in blue slides is not due to a simple acquiescence to the response of
the minority.
1 Thomasand Bistey(1964) reporta studyusingthesamestimulus
as our studyand
theyfoundthat subjectswho called the stimulus"green"or "mostlygreen"showed
significantly
greater
generalization
towardthelongerwavelengththanthosewho calledit
"blue"or "mostlyblue."Our resultsare in the oppositedirection.
2 Usingthesame test,Brownand Lenneberg
(1958) showedthat thereis a relationand color recognition
whichis a functionof stimulusexship betweencolor-naming
since we
posure-time.
Thus we should have varied the exposuretime.Nevertheless
dealtwithhighlycodablecolors,we shouldbe able to recoverthemfromtheirname.But
withtheirswhichshowsthatinconsistency
within
in generalour studyis in agreement
in theindividual.
and hesitation
thegroupcorresponds
to inconsistency

374

SOCIOMETRY

Having raisedthe question:"To whatextentis it possiblefortheseslides


to be perceivedas green"we ascertainedthat subjectsin the experimental
groups did not accept this possibilityin a more significantdegree than
subjects in the controlgroups.On the other hand, however,subjects in
the formergroupsdid prove more inclinedto accept the green response
than subjectsin the lattergroups (t 2.64, p<.008). Thus, we can infer
withtheminority
led to an inclination
thatthedesireto reachan agreement
to see what the latterwere seeing,to make an effortto look for green
in the blue stimuli.With this in mind we asked the subjects: how many
different
nuancesof colordid you distinguish?Subjectsin the experimental
groupsperceivedmorethantwonuances,whilesubjectsin the controlgroups
can also be
saw at mostone or two (Z 2.12, p<.0342). A differentiation
groups.Subjectswho yielded
made betweensubjectswithinthe experimental
to the minoritysay more nuances than those who did not yield to the
minority.(Z=2.79, p<.005). Moreover,whethertheydid or did not yield
to the minoritysubjectsin groupsin whicha changein responseoccurred
perceivedmoreshades than thosein groupswherethe majoritymaintained
its position,and always respondedblue (Z= 1.78, p<.076). Using an appropriatequestion,we thenasked subjectsto specifytheseshadesby naming
the colors which composedthem.No matterwhat these shades were or
how many were cited, for purposesof this analysis we retainedonly the
highestpercentageof green foundon the responsesheet,using it as an
index of the extremelimit of a subject's attemptto find this color. All
subjects in the experimentalgroups distinguishedmore green than those
in the controlgroups (Z=2.99, p<.003). Of course,in the experimental
groups,subjectswho yielded to the minoritysaw more than 30 per cent
(Z- 4.92, p<.001). Everythingtends to point to the fact that members
of the majoritymade an effortto take into account the viewpointof the
minority,
to verifythe objectivebasis of its judgment.At no timedid they
remainpassive,nor were theycontentblindlyto accept or reject a norm
opposed to theirown. The effectof this was probablythe modification,
as we saw, of their own perceptionor their definitionof green and of
blue.
Minority.Naive subjects,whoconstituted
(b) PerceptionoftheConsistent
the majorityin the experimental
groupswere more inclinedto see green
in the blue slides than the controlsubjects (and actually did see more
green). The psychologicalproblemwhichtheyhad to solve was the following: why, althoughhaving agreed that the minority'sanswer was not
did theynot yield to it, since a physicalstimuluswas
withoutfoundation,
was the
involved?The only possible explanationfor such a contradiction

INFLUENCE OF A CONSISTENT MINORITY

375

assertionthat they were less certainthan the minority.Thus while they


wereinterestedby what was proposedto them,theyconsideredthemselves
to be more competentthan the minority,since they representednormal
perception-therefore
theyhad the rightto yield or not to yield. Needless
to say, these trendscan be accountedfor in otherways. In spite of the
resultsof the Polack Test, subjects did not believe that a person who
always perceived these slides as green could have a very good color
perception.Even if he had good vision, his competencyin the area of
colormustbe inferiorto thatof the majorityof people. On the otherhand
the consistentnatureof the minorityresponsein the face of the different
judgmentsemittedby the majority,suppliedgreat self-assurance.
Without
it
can
nonetheless
be
seen
that
the first
comingto any definiteconclusion,
interpretation
applies to the two series of predictionsconsideredtogether,
whilethe secondconcernseach seriesseparately.
Now let us examinethe resultsobtainedmorein detail. In the firsttwo
questionssubjectswereasked to judge each of the personswho participated
in the experiment,
includingthemselves,
on a 10-pointscale (fromgood to
bad), as to theircapacity firstto discriminateintensitiesand second to
perceivecolors.A comparisonof the grades whichsubjectsgave to themand othersubjectsforcolor perceptionis veryinstrucselves,confederates
tive.On thewhole,subjectsconsideredthattheconfederates'
colorperception
was not as good as theirs,both in the groupswherethe confederates
were
seated nextto each other(t=9.98, p<.001), and in the groupswherethey
were separated(t 7.02, p<.001). They also consideredthat confederates
did not perceivecolorsas well the othermembersof the group (t 10.83,
p<.001). Nevertheless,
it was feltthat the secondconfederate
had a better
colorperceptionthanthe firstconfederate(Zz=2.04, p .04, Mann-Whitney
U test). Thus the membersof the majorityjudged themselvesmore competentthantheminority,
and theyexperiencedlittleanxietyregardingtheir
perceptivecapacity.
What about certainty?In theirpostexperimental
questionnairesubjects
had to classify"the personswho participatedin the experiment,
according
to whethertheyweremoreor less sure of theirresponses."Subjectsjudged
to be more sure of theirresponsesthan they were (t-5.02,
confederates
p<.07) and thanothermembersof thegroup(t=4.42, P<.07). A difference
revealeditselfalso in the perceptionof the two confederates.The confederateseated in the firstpositionwas judged as being more sure of his
responsethan the second confederate,
both in the groupswheretheywere
seated next to each other (t=2.54, p<0.7) and in the groupswherethey
were separated (t 3.22, p<.07). These evaluationswere shared by all

376

SOCIOMETRY

subjects,whethertheywereamongthosewho respondedlike the consistent


minority,
or whethertheywerein the groupswherethe majorityresistedall
influence.
Threetrendsclearlyemergefromtheseresults:(a) subjectsjudged
themselvesmore competentand less certainthan confederates;(b) judgmentsof competenceand of certitudeof confederates
had an inverserelation; (c) the confederatein the second positionwas perceiveddifferently
fromthe one in the firstpositionand as being closer to other subjects.
These trendscorroborateobservationsmade in other experiments.
Thus,
Brehmand Lipsher (1958) provedthatperceivedtrustworthiness
would be
greaterwhen the communicator
took an extremepositionon eitherside of
the issue, than when he took a moderateposition.More recently,Eisinger
and Mills (1968) studiedthe effectof the discrepancyof the communicator
positionupon his sincerityand competence.They provedthat a communiand more
catoron the oppositeside will be perceivedas moreincompetent
sincerein comparisonwitha communicator
who is opposedbut moremoderate. These experiments
suggestthat the responseof an individualor an
extremesubgrouphas moreweight.But what interestsus here is the fact
thatobtainingthe same resultsas ours,theyofferindirectsupportin favor
witha normopposed
of the view that consistency,
especiallyof a minority
to the normof the majority,is at the same time an index of extremism.
to the extentthat it shows itselfuncompromising,
Now, this extremism,
and places the othersin
engendersan anxietylinkedto the disagreement,
a situationwheretheymust eitherconcedeor polarize in orderto reduce
and diminishthe anxiety.As nothingpermitsthem to
this disagreement
polarize,thenin certaingroups,subjectsyielded.
The trendsdiscoveredalso enlightenedus about the role of the second
In a sense,he does not contributeany supplementary
confederate.
weight
We make the hyto the responseof the "innovator,"the firstconfederate.
pothesisthat his behaviorservesas an exampleto the othersubjects; he
demonstrates
thatsomeoneis capable of choosingthe minority
responsethat
thereis a choice possible betweenthe two alternativesand to a certain
extent,justifiesthem.In short,if the effectof the firstconfederateis an
influenceeffect,the effectof the second would be what economistscall
effect.In any case the minority'sinfluencecannotbe ata demonstration
tributedto a possible leadershiprecognizedby the group. Questionedas
to which personsin the group they would like to find themselvesin a
morefrequently
similarsituationwith,subjectsdid not chooseconfederates
than any othermemberof the group.Likewise,when asked: "Who would
you like to see lead the discussion(about the experiment)in the group?"
trendcan be observedto choose confederatesless
a slight,nonsignificant
than othernaive subjects.

INFLUENCE OF A CONSISTENT MINORITY

377

DISCUSSION AND CONCLUSION


The experiment
whichwe have just describedshows,at least as far as
female subjects are concerned,that by being consistenta minorityis
capable of influencing
a majorityat the level of verbal and perceptual
responses.But this factmustbe examinedmoreclosely.
GENERALITY OF THE BEHAVIORAL STYLE AS A SOURCE OF INFLUENCE. We
have at the beginningof this articleput forwardthe idea that the consistencyof the behavioris a source of influencewhen a minorityis concernedand whenan innovationprocessis involved.And it clearlyappears
that conformity
is an effectof consistencyand not of dependencetowards
the majorityof the group. To substantiatethis conclusion,we will limit
a groupourselvesto Asch'sexperiments.
We knowthatin theseexperiments
majoritycan induce a single individualto give answersgoing counterto
perceptualevidence.The conditionsrequiredfor this effectto occur are
the usage of a nonambiguousstimulus,the need to respondpublicly,and
the presenceof a unanimousmajority.This majority,accordingto Asch
(1962:497) gives rise to a propensityto adopt the erroneous"conformist"
responsesof the group.Our interpretation
is, of course,different,
but first
let us look to the data and theirmeaning.We can considerthat unanimity
to consistency
in a groupcorresponds
to inter-individual
which
consistency,
resultsfromcoincidenceand identityof responseof several subjects to a
given stimulus.At the same time,the sequence of "erroneousresponses,"
the identityof responsesof each confederatethrougha series of stimuli,
expressesinternal,intra-individual
consistency.What do we see when we
examineAsch's results?We see that a unanimousmajorityfromtwo to
sixteenconfederates
provokedthe acceptenceof "erroneous"responsesfor
one third(32 per cent) of the responsesof the naive subjects.The increase
to more than threehas therefore
in the numberof confederates
no effect
of
these
there
is
no
direct
relationbetween
on thefrequency
responses.Thus,
the magnitudeof this social pressureand conformity.
Now, only one single
confederatein a group made up of seven or eight personshas to break
the unanimityby givingcorrectanswersfor the numberof conformist
responsesto drop to 10.4 per cent or 5.5 per cent. Thus, a groupof three
unanimouspersonsis moreinfluential
thana groupof eightnon-ununanimous
to sayingthatit is the inter-personal
persons.This is tantamount
consistency
of, ratherthan the strengthof social pressurewhich is more important,
and comes closestto accountingfor the variationin the rate of influence.
Asch's (1955) and Allen and Levine's (1968) experimentsgive much
weightto thisinnovation.They thoughtthatif social supportwas important
in order to reduce conformist
constraint,the dissenterought to give the

378

SOCIOMETRY

responsewhich the subjects privatelyconsideredto be correct.On the


was the critical
in the case of unanimitywheregroupconsistency
contrary,
with the group, whetheror not his
variable,a dissenter'sdisagreement
withthe subject'sprivatejudgment,
responseswerecorrectand in agreement
to decreaseconformity.
The resultsof the two experiments
was sufficient
show that it is lack of unanimousconsensuswhichis the decisivefactor.
overtime-of theidentical
ofintra-individual
consistency
Whatis theeffect
repetitionof subjectsresponsesto a series of stimuli?As we know,Asch
responded
used twotypesof trials:"neutral"trialsin whichthe confederates
in a "correct"manner,and "critical" trials in which the confederatesrea groupappearedall
spondedin an "erroneous"manner.Diachronistically,
the moreconsistentwithitselfwhen therewere more"critical"trialsthan
"neutral"ones. Asch (1956) varied the proportionof the neutraltrialsin
relationto the criticaltrials (1/6, 1/2, 1/1, 4/1) and althoughthe difa decrease in the percentageof conformist
ferenceswere not significant,
responseswas observed(50 per cent,36.8 per cent,38.6 per cent,26.2 per
cent) as the majoritybecame less coherentin time. Iscoe and Williams
(1963) obtainedsimilarresults.On the whole,consideringthe information
we have at hand today,we can say that it is the behavioralstyle of a
majorityor a minorityand not the pure amountof social pressurewhich
is revealedto be at the originof influenceexerted.
CHANGE OF VERBAL AND PERCEPTUAL RESPONSES. We have seen that the
alterationof the answer,whilenot negligibleat the conscioussocial level,
at thelatentindividuallevel.Our presentstateof knowledge
is moremarked,
does not enable us to ascertainwhetherit is of a perceptiveor of a verbal
nature (Goldiamond,1958). However,given that most of the experiments
in this field (Tajfel, 1969) with the notableexceptionof Flament (1958)
reportinfluenceat the verbal level and not at the level of perception,
the resultswe have obtainedare all the moreremarkable.They oblige us
betweena changein responseand a changein code, between
to distinguish
at the code level. In thissense,
influenceat the responselevel and influence
in one experiment,
we have the rightto say that the consistentminority,
in the normof the majority,and not only
provokeda real modification
in its response.
If this phenomenonis rare in the laboratory,it is not in politicallife.
Thus, a politicalpartyoftenadopts the ideas or the vocabularyof another
Yet citizenscontinueto vote forthissame party,
partyor social movement.
to respondto this party's slogans. For example,in France the Gaullist
in framingits own educationprogram,adopted part of the
government
the programproposedby studentsand workersin May 1968.
and
rhetoric
Nevertheless,when a Frenchmanvotes for the Gaullistparty he believes

INFLUENCE OF A CONSISTENT MINORITY

379

that he is "responding"to the same political body and in the same


manneras he did in the past, althoughboth it and its representatives
have changedtheiropinionson very specificquestions.Indeed, it is conceivable that minoritiesare more capable of changingthe majority'scode
than its social response,while the majoritywould have more influenceon
the individual'sverbal responsethan on his intellectualor perceptivecode.
This is an historicalreality.Great innovatorshave succeededin imposing
theirideas, theirdiscoveries,
withoutnecessarilyreceivingdirectrecognition
fortheirinfluence.
For example,manypsychologists
have assimilatednotions
elaboratedby psychoanalysis,
all the while refusingto recognizethe value
of psychoanalysis.
it
Thus, if we reallywant to understandthe processof social influence,
and of innovais not enoughto studymorecarefullythe role of minorities
tion.We mustbegin to exploremoresubtle mechanismsof influencethan
those which are at work in direct and visible acceptanceof normsand
judgmentsproposed.
REFERENCES
Allen, V. L.
1965
"Situational factorsin conformity."Advances in ExperimentalSocial Psychology 2:133-175.
Allen,V. L. and J. M. Levine
"Social support, dissent and conformity."Sociometry31(June):138-149.
1968
Asch, S. E.
1955
"Opinions and social pressure."ScientificAmerican 193 (November) :31-35.
1962
Social Psychology. Englewood Cliffs: Prentice-Hall.
Brehm, J. W. and D. Lipsher
"Communicator-communicateediscrepancy and perceived communicator
1959
'trustworthiness'."Journal of Personality 27(June) :352-361.
Brown, R. W. and E. H. Lenneberg
1958
"Studies in linguistic relativity." Pp. 9-18 in Maccoby, Newcomb, and
Hartly (eds.), Readings in Social Psychology,New York: Holt, Rinehart
and Winston.
Eisinger,R. and J. Mills
1968
"Perceptionof the sincerityand competenceof a communicatoras a function
of the extremityof his position." Journalof ExperimentalSocial Psychology
4(April):224-232.
Faucheux, C. and S. Moscovid
"Le style de comportementd'une minorit6et son influencesur les reponses
1967
d'une majorit6." Bulletin du Centre d'Etudes et RecherchesPsychologiques
16(Octobre-Decembre):337-360.
Flament, C.
1958
Influence Sociale et Perception. Annie Psychologique 58(Fasdcule 2):378-

400.

380

SOCIOMETRY

GoIldimond,I. and L. F. Malpasm


Journalof the
1958 "Locus of hypnotically
inducedchangesin colorresponses."
OpticalSocietyof America51(October):1117-1121.
Iscoe,I. and M. S. Williams
behaviorof children."
1963 "Experimental
variablesaffecting
the conformity
JournalofPersonality
31(June):234-246.
Kelley,H. H.
Pp. 192-241in D. Levine (ed.),
1967 "Attribution
theoryin socialpsychology."
on Motivation.
ofNebraskaPress.
NebraskaSymposium
Lincoln:University
McGuire,W.
Social Psy1964 "Inducingresistance
to persuasion."
Advancesin Experimental
chology1:191-229.
Moscovici,S.
1969 BehavioralStyle as a Source of Social Influence.Symposiumon Social
London.
IXth International
Influence,
Congresof Psychology,
Moscovici,S. and C. Faucheux
1969 Social Influence,Conformity
Bias and the Study of ActiveMinorities.
Centerof AdvancedStudyin BehavioralSciencesStrafford
(Mimeo).
Tajfel,H.
Pp. 315-394in Lindzeyand
1969 "Social and culturalfactorsin perception."
Aronson(eds.) The Handbookof social psychology,
Vol. III (2nd ed.),
Reading:AddisonWesley.
Thomas,D. R. and G. Bistey
as a functionof the numberand rangeof gen1964 "Stimulusgeneralization
teststimuli."
eralization
Psychology
68(December):
Journalof Experimental
599-602.
Worell,J.
of exposureto conflict."Progressin Experimental
1967 "Some ramifications
PersonalityResearch4:91-125.

You might also like