Exp eriments in Intergroup Discrimination · 2015-02-23 · by Flenri Tajfel t T t t I I I I t t t T I f; fntergroup discrimination is a feature I of most modern societies. The phe-I

Exp eriments in Intergroup Discrimination

Can cliscrimination be trctced to some suclz origirt as social conflict

ar a history of hostiLity? Not necessarily. Apparently the rnere fact

of division into grolLps is enolLgh ta trigger discriminatory behuvior

by Flenri Tajfel

tTttIIIItttTIf;

fntergroup discrimination is a feature

I of most modern societies. The phe-I nomenon is depressingly similar re-

gardless of the constihrtion of the "in-

group" and of the "outgrouP" that is per-ceived as being somehow different. A

Slovene friend of mine once described to

me the stereotypes-the common traitsattributed to a large human group-thatare applied in his counbry, the richestconstituent republic of Yugoslavia, toimmigrant Bosnians, who come from a

poorer region. Some time later I pre..sented this description to a group of stu-dents at the University of Oxford andasked them to guess by whom it wasused and to whom it referred. The aI-most unanimous reply was that this lvasthe characterization appUed by nativeEnglishmen to "colored" immigrants: .people coming pti*atily from the WestIndies, India and ?akistan.

The.intensity of discrimination variesmore than the natule. of the phenom-enon. In countries with long-standing in-tergroup problems-be they racial as inthe U.S., religious as in Northern Irelandor linguistic-national as in Belgium-ten-sions reach the boiling point more easilythan they do elsewhere. In spite of rliffer-ing econornic, culrural, historical politi-cal and psychological backgrounds, horv-ever, .the ottitudes of prejud.ice towardoutgroups and the belwoior of discrirrii-nation against outgroups ciearly displaya set of common characteristics. Socialscientists have nafurally been concernedto hy to identify these characteristics inan effort to understand. the origins of

. prejudice and discrimination.

Th" invesbigative appronches to this* task can be roughly classiffed into two

categories. Some wor.kers sbess the so-cial determinants of preiudice and, di3-criminatioa. Others emphasize psycho-

' 9 6

logical causation. In The Functions ofSocial Conf.,ict; published in 1956, LewisA. Coser of Brandeis University estab-lished a reiated dichotomy when he dis-tinguished behveen two types of inter-group confl.ict: the "rational" and the "ir-rational." The forsrer is a means to anend: the con$ict and the attitudes thatgo with it rbflect a genuine competitionbetween $oups. with divergent interes ts.The latter is an end in itself: it serves torelease accumulated emotional tensionsof various kinds. As both popular loreand. the psychological literah:re testify,nothing is better suited for this purposethan a well-selected, scapegoat.

These dichotomies have some value asanalytical tools but they need not be

. taken too seriously.'Most cases of con-flict between human groups, large orsmall; . reflect an intricate interdepen-

'dence of soiial and psychological cau-sation. Often it is difficulg and probablyfruitless, to speculate about what werethe ffrst causes of real present-day socialsih:ations. Moreover, there is a dialecti-cal relation between the objective andthe subjective determinauts of intergroupabtitudes and behavior. Once the processis set in motion they reinforce each otherin a relentless spiral in wbich the weighto[ predominant causes tends to shift con-tinirousiy. For example, economic or so-cial competition can.lead to discrimina-tory behavior; that behavior can then ina number of ways create attitudes bfprejudice; those attitudes can in turn'lead to new forms of discriminatory be-havior tbat create new economic or so-.cial disparities, and so the vieious cireleis continued.

The interdependence of the two tyllesof causation does not manifest itself onlyin their mutual reinforcement. They ac-to"ily converge because of the psycho-logical effects on an individual of his so-

ciocultural milieu. This convergence isoften considered in terms of socia! Ie:rrn-ing and conformity. For instance, there ismuch evidence that children learn quiteearly the pecking order of evaluations ofvarious groups that prevails in their so-cief, and that the order remains fairlystable. This applies not only to the evalu-ation of groups tliat are in daily contaclsuch as racial grouPs in mixed environ-ments, but also to ideas about foreignnations with which there is Jittle if *ypersonal contact.

fn sfudies conducted at Oxford a fervyears ago rny colleagues and I found ahigh consensus alnong children of sixand seven in their preference for'fourforeign counhies. The order was'Amer-ica, France, Germany and Russia, andthere was a correlation of .98 betweenthe preferences of subjects from tlvodi#s1ss1 schools. As for adults, studiesconducted by Ttromas F. Pettigrerv inthe late 1950's in South AJrica and in theAmerican South have shown tbat con-formity is an important determinant ofhostile attitudes toward blacl<s in bothplaces (above and beyond individu-al tendencieb toward. authoritarianism,which is lcaown to be closely related to

prejudice tcilvard outgrouPs).

rfrhese stud.ies, Iike many others, werer concerned with attitudes rather than

behavior, ivith prejud.ice rather than dis-crimination. Discriminatioq it is oftensaid is more {irectly a function of theobjective social situation, which some-times does and sometimes does not fa-

cilitate the erpression of attitudes; theattihrdes of prejudice may be sociallyleamed, or due to tendencies to conform,but they are not avery efficient predictoro[ d.iscriminatory behavior. According to

this view, psychological considerationsare best suited to explaining and predict-

A

MATRIX 1

MATRIX 2

B

MATR1X 3

MATRIX 4

C

MATRIX 5

MATRIX 6

FIRST EXPERII{ENT conducterl by the aurhor anil his colleaguesutilized these six matricee. The nnr',hers representetl points (Iater

trairslated into awards or penalties i.E money) to be assigned by asubject to other inilividuals; bycheclcing a box the srirlject assigneilthe nunber of points iu the top of the box to oue persou anil thenumber in the bottom ol the box to ,nother person.; he tl id notkoow the identiry of these people but only whether each was amember of his owu group or "the other group." (The groups had

been established by the experimenrers oh grounils tlat-were arti.ficial ancl ineignificant.) Each reatrix appeareil three ti-es in a testbooklet with each row of nurnbers labeled to inilicate wbether tLesubject was choosing berween two members oI his own group (in-group) otber.than himseU, two membere of the oritgroup or oue

. me'.,her of the ingroup and one member oI the outgroup. Choiceslyere scored to see if subjects chose'for fairness, maximu,m gain totheir owu group or msximum difierence in favor of the ingroup.

- 1 9 - 1 6 - 1 3 - 1 0 -7 -4 -1 0 IT 2 3 4 5 6

o 4 1 2 .tI 0 I -4 - 7 - 1 0 - t J - 1 6 -19

12 10 I b + 2 n I- l -5 -9 - t J -17 -21 -25

-25 -21 - 1 7 - 1 3 - 3 I- l n 2 4 h t' I \ J 12

I 2 4 h 6 7 I I l v t l 12 1 3 1 4

14 1 3 1.2 t l 1 0 I I I A 5 4 3 2 1I

1 8 17 16 1 5 1 4 1 3 12 11 1 0 :t I 7 6 5

5 6 7 8 I t [ l 11 12 1 3 1 4 1 5 16 17 18

-14 -12 - 1 0 -8 -6 -4 -2 - 1 c 1I 11 l c I Y 23

23 { nI J l c 11

-lt 3 I -2 -4 -6 -8 - 1 0 -12 -14

.4 ''I I 1 4 .l .l

t l I 2 - 1 -2 -4 -5 -o .'_ T -8

-8 -7 -6 -5 -4 -3 -2 I 2 I il 1 4 1 7

tI

ing the genesis and functioning of atti-rudes; the trrcts of intergroup discriminn-tion are best related to, and predictedfrom, objective indexes o[ a social, eco-nomic and demograPhic nafure.

Although l ltave no quarrel lvith thisvierv; I am leFt with a nagging feelingthat it omits an important part of thestory. The fact is that behavior torvard.outgroups shorvs the same monotonoussimilnrity a.s rrttitudes do, across a diver-sity of socioeconomic conditions. Thisapparent diversity may, of course, ob-scure an underlying common firctor of"rirtional" confict, of struggle to pre-sele a sfattrs quo favorable to oneselfor to obtdn an equitable share of socialopportunities and beneEts. Another kindof underlying regularity is nonethelesscommon to a varieby of social sifuationsand is an important psychological 9ffecto[ our socioculhual mi[eu. It is tbe as-similation by the individual of tfie vari-ous norms of conduct that prevail in hissociety. .

Etot the prrrposes of this article I shallr define social norms as being an in-dividual's expectation of hori otherserpect him to behave and his expectationof how others will behave in any givensobial situation. Whether he does or doesnot behave according to these expecta-

MAXIMUM GAINFOR OUTGROUP

MEMBER

MAXIMUM GAINFOR INGROUP

MEMBER X

tions depends primnrily on his under-standing of lvhetirer or not and horv asituation relates to a specific set of ex-pectntions. If a link is made between theone ilnd the other-if an inclividual's un-derstnnding o[ a situation in rvhich heffnds himself is such that in his vierv cer-tain frrmiliar social norns ale pertiuentto it-he behaves accordingly.

There is nothing nerv to this formula-tion; it is inherent in most studies anddiscussions of intergroup prejudice anddiscrimination that stress the importanceo[ conforrniry. The p.oint I wish to ma]<eis broader. Conformily contributes tohostile attitudes and behavior torvardspecified goups of people in situationsthat are usually characterized by a his-tory of intergr'oup tensions, conf.icts ofinterest and early acquisition by individ-uals of hostile vielvs about selected out-groups. lVe are. dealing, however, lvitha pr:ocess that is more general and goesdeeper than the learning of value jrrdg-ments about a speci-Ec group and thesubsequent aeting out of accepted pl,t-terns of behavior toward that group. Thechild learns not only rvhom he shouldIike or dislike in the complex social en-vironment'to lvhich he is exposed butalso something more basic. An infividualconsb'ucts his orvrr "web of social afi]ia-tioni" by oppllong'principles of order

INGROUP-OUTGROUP CHOICESr*l

INGROUP- tNGROUP CHOICES

OUTGROUP.OUTGROUP CHOICES

and simplification that reduce the com-ple.riby of crisscrossing human categori-zations.Perhap.s the most important prin-ciple of the subiective social order rveconshuct for ourselves is the classifica-tion of groups as 'we" and "they"-as in-groups (any number of them to rvhichrve happen to beiong) and outgroups.The criteria for these assignments mayvary according to the situation, and theiremotional impact may be high or lorv,but in. our socielies this division intogoups most often implies a competitiverelation behveen the groups. In otherrvords; intergoup categorizations of allkinds may bring into play what seems tothe in&vidual to be the appropriate formof intergroup behavior.

What this essentially means is that theneed to bring some kind of order into our"social conshuction of reality" (a termrecently used by Peter L. Berger of theNerv School for Socia] Research andThomas Luckmann of the University ofFranldurt) combines with the hostilityinherent in many of the intergoup cate-gorizations to which we are continuallyexposed to develop a "generic norm" ofbehavior torvard outgroups. Wheneverwe are confronted with a situation towhich some form of intergroup categorization appears directly relevant, we arelikely to act in a manner that discrimi-

MAXIMUM GAINFOR INGROUPMEMBER

M,AXIMUM GAINFOR INGROUPMEMBER Y

tItIITI]

;

ttItItIIx

MAXTMUM GAINFOR OUTGROUP

MEMBER X

RXSWTS \9ERE SCORED by ranlcins the choices from I to 14ilepending on which bor was checkeil. The encl of the matrix atwhich the ingroup member got the minimum number of points(and the outgroup member tbe maximum) was clesignatecl l; the.otber end, giving the ingroup member the maximum, was 14, The'

9Btr

,1 .

MAXIMUM GAINFOR OUTGROUPMEMBER Y

mean choicee (colored uertical lines) are showa Lere" In the inter.group situation thb subjects gav€ significantly rnore points to tnem-bers oI their own group than to members of tLe'other group.In theintragroup nituations, however, the means.ol the choices.fell at.Rank ?.5, betwe'en the choices of maximum fairnese (brackets),

\

\ uI

I/ )

\\ / ,

)

\

\ hI

50

40

Fzlrl

trt r I

2n| / ) J v

UJO

I()LL

( J c v

ztlJ

l

IJ-JEtJ-

' '10

0

MAXIMUMFAIRNESS

5 6

MAXIMUMDIFFERENTIATION

nates agAinst the outgroup and favorsthe ingroup,

If ttr.is is true, if drere exists such ageneric norm of behavior toward out-groups, several impoltant consequencesshould follow. The ffrst is that there rnaybe discrimination ag&inst an outgroupeven if there is no reosott for it in termsof th.e in&vidual's orvo iirterests-interms of what he can gain as a result ofdiscriminating against the outgroup. Thesecond consequence is that there may besuch discrimination in the ibsence ofany previously eristing attitudes of hos-tility of disUke torvard the outgroup. Andthe third consequence, following direct-Iy from the secbnd, is that this genericnorm may manifest itself directly in be-havior toward the outgroup before anyattitudes of prejudice or hostility havebeen fo-rmed. If this reasoning is correct,then discriminitory intergroup behaviorcan sometimes be e*pect"d Jven if theindividiral is not involved in ach-ral (oreven imagined) genfliqts of interest andhas no past history of attitudes of inter-group hostility.

[t the Universily of Bristol, in collab-a ! oration with Claude Flament of theUniversify of Alx-Marseille, R. P. Bundyand M. J.Billig, I.have conducted exper-iments designed to test this predictionand others that follorv from it. The mainproblem lvas to create experimental con-didons that would enable w to assess theeffects of intergroup categorization perse, uncontaminxtsd by other variables,such as interactions'among ind.ividuals or

. preexisting attitudes. We aimed, more-over, to look at the behavior rather thanthe attitudbs of the subjects toward theirown group aod the obher group, to en-sure that this behavior rvas of some im-portance to them and to present themrvith a clear alterirative to discriminatingagainst the outgroup that rvould be amore "sensible" mode of behavior.

Perhaps the best means of conveyingthe way these criteria were met is to de-scribe.the procedure we followed in thefi.rst erperiments and its variants in sub-rrqt rrit ones. Our subjects wele 64 boys14 and 15 years old from'a stare, or"comprehensive," school in a subru.b ofBristol. They came to the laboratory inseparate goups of eight. AII the boys ineach of the groups were from the samehouse in the sarne forrn at the school, sothat the1, knew each other well beforethe erperirnent. The fust part of the e.r-periment served. to establish an inter-group categbrization and in the secondpart we assessed the'effects of that cate-gorization on intergroup behavior.

INTERGROUP DISCRIMINATfON was a deliberate strategy in the ingroup.ourgroupchoices (colored' curue) and fairness a deliberate strategy in the ingrtiup-ingrou.p (Cra.yland outgroup'outgroup (btaclc) choices. This is inclicated by the fact that the frequencies ofintergroup choices tlifierecl significandy from those oI the intragroup choices only at theexlreme points of the distribution, the points oI ma.',cimum fairness ancl of maximum dis.crimination. (For this analysis the tw:o lairest choices in each matrix, the two middle ones,lvsJs lnnksd together as 0 and deparlures in either d.irection were Ecored lrom 1 to 6.)

Inthe firstpart tl-re boys were broughttogether in a lecture room and were toldthat. we were interested in the study ofvisual judgments. Forty clusters of vary-ing numbers of dots were flashed on ascreen. The boys were asked to estimatethe number of dots in each ciuster and torecord 'eacir estimate in succession onpr:eparecl score sheets. There wer'd trvoconclitions in this first part of the experi-ment. In one condition, aftei the boyshad completed their estimates they weretold that in judgments of this kind somepeople consistently overestimate thenumber of dots and some consistentlyunderestimate the number, but thatthese tendeneies are in no way relatecl toaccuracy. In the other condiUon the boyswere told that some people are consist-entiy. more accurate than others. Forugroups o[ eight servec] in eac]r of t]re'trvoconditions.

After the judgments had been madeand ha.d been ostentatiousiy "scored" byone of the experimenters, we told. thesubjects that, since lve lvere also intei:ested in other lcinds of decision, lve \veregoing to take advantage of their.presenceto-investigate tliese as lvelf and that for

ease of c.oding.lve lvere going to groupthem on the basis of the visual iudg-ments they had just made. In actualitythe subjects rvere assigned to groupsquite at rand.om, half to "underestima-tors" and half to "overestimators" in thefirst condition, half to "better" and halfto "worse" accuracy in the second one.

Irubuctions follolved tr,bout the nabureof the forthcoming task. The boys.weretold that it would cousist of giving toothers relvards and penalties in real mon-ey. They would not know the identity ofthe individuals to whom they would beassigning these rewards.'and penaltiessince everyone would have a code num-

' ber. They rvould be taken to anothelroom one by one and given in-formationas to which group they lvere in. Once inthe other room they were to lvork bntheir own in sepalate cubicl,es. In eac.hcubicle they rvould find a pencil and abooklet conbaining 18 sets of orderednumbers, one to . each page, It wasstressed that on no occasion rvould theboys be relvalding or penalizing them-selves; they lvould always be allottingmoney to others. At the end of tle.taskeach boy would be brought back into

99

A

MATRIX 1

M I P

. M D

MATRIX 2

B

MATRIX 3 M I P

M J P

M I P

M J P

M D

MD

MATRIX 4

SEcoNDEXPERIMENTinvo lve i l newmat r i ces .Eachwas .p re .

sented in four versions labelecL (as in rhe illustratiou at the bottom

of this page) to inclicate whether the choice was between membeis

of ilifferent groups or berween two members of the sane Eroup; '

the intergroup choices sometin'es bact the ingroup member'g points

in the top row "rra

.o-tti-es hacl them in the bottom row'' The ob'

jective nolv was to *"1y"" the influence of three variables on t'he

sublects' choices: m'aximum ingroup profit (MIP)' maximum joint

;;;, (MIP') and maximum difierence in lavor of the ingroup

member (MD). These varieil according to ili'frerent patterns in the

Type L and Type B matrices ancl in the ilifferent versions; in sorire

cases the maximg were together at one end of the matrix antl in

other cases they ln."" "i

opiosit" eucls' Fdr' example' in the ingroup'

over-outgroop t...iore oiiyp" 'C matrices the maximr:m ingroup

profit and maximum ilifierence were at oue enil antl the maximum

ioilet pront.at the other end' (colored twe\; in the oEtgroup'over-

ingroup version oi tl" s"-e lltrices the three maxima \YeEe to'

.gether at the right'hand encl of the mauiees (blaclc Wpe\' Type B

ingroup'ov".'oo,*ti-tip o"ttioot' -on the bther hand' distinguish the

diference in favor- oi i"g'otrp from the other two gains (color)'

M J P

MJP

M I P

M D

20 1 9 1 8 1 1l . l 16 1 q 1 A

t -1 q 12 11

23 22 21

1 t r i 7 1 9 21 23 25 21 29q 7 I { l

l t 13

1 4 1 5 1 6 17 1 8 1 q

7 I I 1 0 11 1 ? 1 3

7 I 11 1 3 1 5 a 1l , 1 9 21 23 25

I

20 21 22 23{ . tt l 12 1 3 1 4 15 16 1 7 1 8 1g

21 23 25 27 29J 1 o 1 1

l l 13 1 5 1 7 19

PAGE OF BOOKLET, presentiug a single matrix' is reproduced as

a subject might have m".kecl it. In addition to checking a box' the

subject fi i leil ' in the blanks below it to confirm his choice' The page

r00

heacling reminiled him which group he was in' The awarils rver

made to persons itlentified only by number and group; the subjet

'icr nor know ,"h"-;il;;-r"".u Loi onry their group iclenrificatior

the fust room and would receive theamount of money the other boys hadawarded him. The value o.[ each pointthey were awarding tvas a tenth of apenny (about a tenth of a U.S. ceni).Afber these instructions were given, tleboys were led individually to tireir cubi-cles to fill out their booklets.

On each page in the booklet there wasone makix consisting of 14 boxes con-taining two numbers each. The numbers'in the top rolv were the rewards andpenalties to.be awarded to one personand those in the bottom row rffere thoseto be awarded to anolher. Each row waslabeled "These are relvards and penaltiesfor menber No. - of your group" o,"...of the other group." The. subjectshad to in&cate their choices by check-ing one box in each mabix. On the coverof each booklet and at the top of eachpage was written: "Booklet for memberof the group."

fhere were six mahices [see itlust a-L tion on page g7] and each of them

appeared three times in the booklet-.once for each of three $pes of choice.There wer.e ingroup choi.s, with thetop and the bottom row signifying therewards and penalties to be awarded totwo members of the subject's own group(other than.himself). Then there wereoutgroup choices, with both rows signi-fy=og the rewards and penalties for amember of the other goup. Finallythere were intergroup, or "r{i$grentia!"ehoices, one row indicating the rewardsand penalties to be awarded to an in-group member (other than himself) andthe other. the points for an outgroupmember. (The top and bobtom positio*of ingroup and outgroup members werevaried at random.)

The results for the intergroup choiceswere first scored in terms of ranks ofchoices. In each matrix Rank 1 stood forthe choice of the tenn that gave to themember of the ingroup the" minimum

. possible nurnber of points in that matrix;RanI< 14, at the opposite exbeme of thematrix,'stood for the maximu:n possiblenumber of points. Comparable (butmorecomplex) methods of scoring were adopt-ed for the other two kinds of choice, theingroup choices aad the outgroup ones,and for comparison of these choices withthose made in the differential situation.

The results ]vere sbiking. In makingtheir intergroup choices a large majorityof the subjects, in ail groups in both con-ditions, gave more money to members oftheir own group than to members of theother group. All the results were-at avery high level of statistical sign;fig*r..

-above both Rank 7.S, which representsthe point of ma-xjmum fairness, La tn"mean ranla of the ingogp and outgroupchoices. ln conbast the ingro,rp *nI oot-group _choices .tvere closely &sbibutedabou[ the point of fairuess. F,rrth", anal-ysis made it elear that intergroup dis_crimination was the delibera6 sdategyadopted in making intergroup choices.

Before continuing, Iet us revieW thesituation. The boys, who knew each oth_-er well, were djvided into groups definedby flimsy and unimportauicritJri*. Th"irown individual interests were not affect-ed by_their choices, since they aiways as-signed points to two other peopie andno one could know what any othlr bpy,,choices rvere. The amounL of monevwere not trivial for them: each boy leftthe experiment with tle equivalerit ofabout a dollar. Inasmuch

", thuy.couid

not know who was in their group andwho was in the other group, Ihey'couldhave adopted either oT t o reasonablesbategies. They could have,chosen thema..cimum-joint-profit point of the ma-.trices, which would mean.that the boysas a total group would get the most mon-.ey 9_ut qf the experimenters, .or theycould choos'e the pour*t of ma-ximum fair-ness. Indeed, they did tend to choosethe second alternativewhen their choicesdid not involve a distinction bebween in_group and outgroup. As soon as this dif_ferentiatiotr was involved, however, theydiscriminated.in favor of the ingroup.The only thing we needed to ?o toachieve this lssuh was to associate theirjudgments of numbers of dots witd.theuse of the terms "yo* Soup,' and ..theother group" iu. the insU.uctiins and onthe booklets of matrices.

he results lvere at a very high levelof statistical

_significaoce in alt eight

lfParately-gted groups of eight boys.In vierv of the consistency of the phe-nomenon we.decided to analyze it-fur-ther and also to validate it \\rith a d.iffer-

subsequently itr their individual cubicleswere different from those in the ffrst ex-periment. We were norv interested in as_sessing the relative rveights cjf some ofthe variables that may have pulled theirdecisions in one dtection or the other. Inthis experiment lve Iooked at three vari-abies: ma.timum joint profit, or the larg_est possible joint award to both peopl!;ma-rimum ingryup profit or the largestpossible arvard to a mernber of thJin-groupr and ma.timum difference, or thelargest posSible difference in gain be-tween a member of the ingroup and amember of the outgroup inlavor of theformer. ."

There were' .four difierent mabiceslsee tep illustration on o.pposite pagel.As'in the first experiment, there werethree tfp"r of choice: between a mem-ber of the ingroup and a member of theoutgroup, between tr,vo members of the.ingroup and bebween two members ofthe outgroup. In the outgroup-over-in-gFoup version of Type A mabices (thatis, where the numbers in the'top rowrepresented amounts gtven to a memberof the outgroup and in the bottom rowto a member of the ingroup) th" tlireegains*jointprofit, ingroup profit and dif-ference in favor of the ingroup-variedtogether; their maxima (mlxr.mum jointprofft, ma.rimum ingroup profft andmanmum difference) were all at thesame end of the matrix. In the ingroup-over-outgroup version, ingroup profitand difference in favor of ingroup wenttogether in one direction and were indirect ssnflisl with choices approachingrna-rcimum joint profit. In the Type Bma bices outgro up-over-in goup versionsagain represented a covariation of thethree gain.1; in the ingroup-over-out-group versiod,s, difference in favor ofingroup varied in the direction oppositeto joint profit and ingroup profit com-bined.-

A comparison of the boys' choices inthe various mabices showed that ma-ri.mum joint profit exerted hardly any ef-fect at aII; the efiect of mzuCmum in-gloup profft and ma.rimr::n difierencecombined against madmum joint profftwe,s shong and highly significant; theeffect of maximum difierence against-ma*''cimu:n joint piofit and ma..cim,ril i;-group profit was also sbong and highlysignificant. In other lvords, when-thlsubjects had a ihoice between marimjz-i"99"_ profit f91 ali and maximizing thep_rofit formem,bers_cif their own group,ttrey acted on behalf of their o* [-oup.When they had a choice between profttfor all and for tleir own group lo*-bined, as again.sl their own gtrOup's win-

ent criterion for intergroup categoriza-tion, We tested three i"*ioops"of 16-.boys each,

-this_ time with ae-stheic pref_erence as the basis of the division into .t_wo gioups. The boys were shown 12slides, slx of which wete reprod,uctions

ll pT"$gs -by-Paul Klee i.nd sir byWassiiy Kandinsky, and they were askedto erpress their preference for one or theother of these two "foreign pejnlg15.,,The reproductions were prJren*ted. with_out the painter's signature, so that ha,lfof the subjects could be assigned, at ran-dom to the "KIee group" *a n*U to the"Kandin5\, group.,'

The matrices &at confronted the boys

I 0 t

fow'luE,. \Sf ,

Oxfo r dtIIIIffiIIIiiirir:

IIIIIffiIIIItr

The.Face of the DeePBv BRUCE C- HEEZEN' f'gmonrDolt'erty

{;l:"i#ip*#*:"n[llfxq'#i[;;"id";[i[i. or the ocean botto-m in more

than 600 photographs, eight of.which 1T"' inlrtl'ri'606' iT" ioii"pr's, "igh t of. which are in

)".1i^.'t.ft ani ci,ptions"provide a full de':,;i;." t""fi ' -""

a" - "ip

tif ns" p r o vid e I f:]l dl'lliiJ;r"i.";i iil -""*tv uriexplored. area' In.il; ;il ;t " r.s u.ch f e a tu' l t "t .tLt -{ H:9:-t-1.,'.'i i las;- ilrUidity. currents, bottom currents'

il;";L G;i t'"6ks snd tpirs'- 1!1 l-**::"t:'iit"i,ttto"a trails, the autho-rs p.re'

[iTt ;ii;"fii ;"-;i;b I ; ; " d c u r r e n ! i n t r o d u c t i o n

i " ' ohloi?"t 'o ceanogtaPhY' 'i;iJ,;i'"1;ii 6so p'p.'odai l lus. cloth $25.00

paPer $12.50

Exploring the UniverseSeconcl Edition,Editeclbr LOUISE B' Y0UNG' Scicnce Etlitor'

i;;' ;;; ; i,, - i oii-ai' tio

" fo r C o ntin u i n g Ed u ca'

i i r t . ' F.." i , t i , ,g t t t or i6ntat ion for an. under'

rr"ilai.* 'oi

tl 't. principles upon rvhich the

:;il' ilt i "''il.t'i u "i lt,

th is. b'o o n,o r':.1i lnq?i [ r , t" t"tuggests the methods andtature of

"t"";t L.prE"tation oi the universe' T-he reader

il'i ltA';;"i, tn.Gh the .use o[ orieinal

;rt , t ; ; : ; ;- i [e-atr i t ' iaes and moti*at iois of

scientists rangrng-from Nervtor.r and .his dis'

;ili;;"i-n? 'Er'ottion of light to the mod'

ffi";;;;;l"si." and their dibates over the

ii".ii* "f

cieation' Emphasis is..placed uoon

;ii;"reil;r. oi =ti"nt" to art' philbsophv' irnd

Hi:i:,!l""OtI 6so pp. 2oo llus.cloth $12.00paper $6.50

Br EVA NOVOTNY, StaffScien'tist' Astronorny

hi;;i:' Ivlanned ftiace{ruft Qentgr' .Houston'"\i;i;;;

"i-itt" intdrmediate level' this intro'

i;;,i;; to -".ttopttysic-s

explores the basic

In*.tti""f "on""ptJ

oI steUar atmospheres

;',"f i;i*;i"rs- It iricludes such.topics as radta'

;i;; "

,;;;;i"i -it

"o.v' model atmospheres'

il; ; d ii " ;:i i " " f o' * i t i o l:,,"-1"191, fl I l'i' ̂ t i o''

An Introcluction to Stellar

Atmospheres and Interiors

q u r e ' r - ' v - - ' - n r l s t e l l a r e v b l u t i o n 'decenera te matLer 'degenera l

Fsll I971 approx. 350 PP.

CXFORD WUNIVERSITY }ry

PRESS }g200 Madison Avcnue,

Ncw York, N.Y. 100.16

Nlolecuiar Q uantum Mechanics :

An Introduction to

Quanturn Clt emistrY

Bv P.W. ATKINS' Oxford Uniuersily' This

b6ok introduces the ideas--ot. guantum me'

"fr"^r". '" i .-P; ' t l . Part I I dei ls with the

;;'ib; ";

i;oi ro " "d

o t i ons-o f th e-th e o ry, wh ile

ii;;'l i i-;ft'ers an account of aroniic and

*otl."t"t structure and spectra and further

;;-r;;il; eiectric and maghetic. properties of

molecules. The rvork is available as a tlYo'

""f"-rn- p"perbouird set, or in one volume'

c lo thbound, $ I7-?5.i'9i0 V;:'i '1P*1t',I, il1 230 pp' VoL iI (Part

n4 25i PP. Pi let, each $5'50

102

them.Evidence leading in the same firec'

tion emergecl from the other hvo types of

ehoice, bJrween two rnembers of the in-

group and. behveen tlvo members of the

iutgioup, the ingroup choices were eon-

sistJndy and significantly nearer. to tire

madmum joint proftt than were the out-

group ,.rui-"o& this lvas so in spite of

ifr" tn"t that giving as much as possible

to trvo.memb-ers of th" out$oup in the

choices aPPlylng solely to them Pre-sented. tto *ttfli.t with the ingroup's in-

ning more than . the outgroup at -the

,*.iifi." o[ both of these utilitarian ad-

vantages, it rvas the ma'ximization o[ dif-

ferenie that seemed more important to

terest. It simply would have meant giv-

ing more to ithe others" lvithout gt"t"g

*.r] l"rt to "yout olvr." This represented"

therefore, a clear case of grahritous d'is-

crtmination. We also included in the

s'econd. dxperiment;om.9 of the original

matrices used in the firit one, rvith re-

sults much the same as before' Again aii

were asked to do. It was enough for them

to see themselves as clearly categorized'

into an ingroup and' an oulgroup, flimsy

as the criteria for this division were-

even though the boys knew one another

lvell before the erperiments, tierr orvn

individual gains oi.te not involved in

their decisions and their actions could

have been aimed' to achieve the gi-eatest

corunon good.

Tt rvould' seem' then, that the genericI- oo* of outgroup behavior !o which

I have referrel dols exist and that it

helps to distort what might have been

;;;. reasonable conduct' This norm de-

termines behavior-as ot'her social norns

do-rvhen an in&vid'ual ffnds himself in'

a situatioir to which, in his view' the

norrn applies. Behavior is never- motive-

less, bul it is a crude oversimphffcation

to ttink that motives in social sihrations

include no more than calculations of

seif-interest or that they can be derived

from a felv supposedly universal human

drives such as aggression toward the out-

sider, the need to "nUote

and so on: To

behave socially is a complex business' It

involves a iong learning process; it is

based on the i:anipulation of qymbols

and. abstractions; it implies the capaci$

for mociification of cJnduet when the

situation changes-and social situations

never remd.in static. To behave aPPro'

priately is therefore z pow-erful social

motive, and attemPting to do so rheans

to'behave accord-ing to one's best under-

standing of the situation.-Judg*ents of

rvhat dappropriate are determined by

social ttot*s, or sets of axpectations'

It seems clear that two such norms

were understood by our subjects to apply .to the situation we'imposed' on them:

'

"groupness" and "fairness." Thgy man-

o!.d io achieve a neat balance between

ti'. t*o, and one might asfllme that in

real-life sihrabions the same ldnd of

balance would aPPIy. Unfortunately it is

only too easy to ;hdk of examples in real

life nuhere fiairness would go out the

wind.otv, since grouPness is often based

on. criterift more weighty'thau either

preferring a painter ooi h*t never heard

of b"fo.. or iesembling someone else in

one's lvay of counting dots' So-cialization

into "groupness" is powerful and un-

avoidaile; it .has innumerabie valuable

functiotu. Ii also has some odd' side ef-

fects that may-and do-reinforce acute

intergroup tensioirs whose roots lie else-

ruheti. Plrhaps those educators in oul

compehitive societies who from the ear]i-

est sihooling are so keen on "teams" andir"** spiril could give-som-e thought to

the opeiation of these side efiects'

the results in this erperirnent were at a

high level of statistical significance'in subsequent experiments lve tested

the importance of fairness in making the

choicei, the effect on the choices of

familiarity with the situation and the

subiects' ideas about the choices that

othe.s lvere making. Fairness, lve fourrd,

was an important determinant;'most of

the choices must be understood as being

a comPromise between fairne-ss ancl,fa-

| "ori"g one's olvn groug' lVe fgun{ 1ha1

$9.s0

discriirittation not only persistecl but

also increased when the entire situation

became more familiar to the subjects'

lvith familiarity there was also an in-

crease (lvhen the boys lvere asked to

pred.ict the other subjects' beiravior) in

iheit'erpectabion that other boys rvere

fiscriminating.Much ,"*oitts to be done to analYze

the entire phenomenou in greater detail

and to gain a fuller understanding of its

determining conditions, but some clear

inferences can ah'eady be made' Out-

group discrimination is exlraordinarily

easy-to trigger ofi. In some-previous

sfudies of' 'group conflict, such ns one

conducted 6y U,rz"ter Sherif at tbe Uni-

versif of Oklahoma, grouPs had to be

placed in interue competition .for sev-

lral days for such results to opcur lsee"Erperiments in Group Con-0'ict " by

lv{uiafer Sherif; ScrsNTrrrc Arvruntce'N,

November, 1956]; in other situations

behavior of this kind can occur without

direct confl.ict'if it is based on previously

existing hostility, Yet neither an objective

conflict o[ intelests nor hostility had any

relevance rvhatever to lvhat our subiects