-11D 091 796

AUTHORTITLEINSTITUTIONPUB DATENOTEAVAILABLE FROM

JOURNAL CIT

EDRS PRICEDESCRIPTORS

DOCUMENT RESUME

CS 500 722

Bowers, 4ohn Waite, Ed.Communication and Conflict.Speech Communication Association, New York, N.Y.Mar 7496p.Speech Communication Association, Statler HiltonHotel, New York, New York 10001 ($5.00)Speech Monographs; v41 n1 Entire Issue March 1974

MF-$0.75 HC-$4.20 PLUS POSTAGE*Communication (Thought Transfer); *Conflict;Conflict Resolution; Decision Making; Dissent; GroupRelations; Higher Education; Nonverbal Communication;Problem Solving; Theories; Verbal Communication

ABSTRACTPapers in this special issue of "Speech Monographs"

focus on the communicative aspects of conflict as an important butneglected area of research. John Waite Bowers introduces thepublication with "Beyond Threats and Promises," while David W.Johnson analyzes conflict literature in the longest paper,"Communication and the Inducement of cooperative Behavior inConflicts: A Critical Review." Additional papers include thefollowing: "An Analytic Model of Conflict" by Charles E. Watkins;"Conciliation and Verbal Ref:Oases as Functions of Orientation andThreat in Group Interaction" by Theodore Jon Nam "The.EffeCts ofSubstantive and Affective Conflict in.Probles-Solving Groups" by MaeArnold Bell; "Communication in Game Simulated ConfliCtst TwoExperiments" by Thomas M. Steinfatt, David R. Seibold, and Jerry IceFrye; "Argurent in Negotiations A Theoretical and Empirical Approach"by Nancy A. Reiches and Harriet B. Harral; "Perceiving.CosMunicatiOUConflict" by Thomas J. Saine; and "A Literary Analog to ConflictTheories: The Potential for Theory Construction" by Lawrence J. Chaseand Charles N. Kneupper. (JM)

V I of AllameNtoir meAtI4,I Ou0TioN wiLlAtiNATioNAL INstituTE OP

EDUCATIONTHIS DOCuMENT HAS BEEN REPRODuCED EXACTLY AS RECEIVED PROMTHE PERSON OR ORGANI1AriONORIGtAiMING IT POINTS OF VIEW OR OPINIONSSTAIEO DO NOT NECESSARILY REPRESENT OFFICIAL NATIONAL. INSTITUTE OFEOUcATioN POSITION OR POLICY

IP

SPEECHMONOGRAPHS

1'U1511511E1) .11Y

'1111: Sl'EE.C11 CONIMUNICA*111ON ASSOCIr1tiUN

(V{IIIAA1 WORK, (..%51 Mitt: ,,N:( tvl.i1ySlatIck (WWII IButCI. Ntsv Yolk (Its 10001

CDITOR

T110,NIAS M. SCIIWELDepartment of C:onuntillicationAtts

University of 'iscotisinNIadistrit, NVisconsin 53706

EDITORIAL ASSISTANT

KAY AL DEN I3RACF.University of Wisconsin

ASSOLIA I ,1,1 1/1

1111 IIoNn N. III N5:11'1 I, 0115 eisil5 ur 11',Oqugluil(;soRrii nokois, 1't:11115,05mila SIMI: ()OA cl silyROM al I). lik040).5, Not am Este! II (lulu:1,51(yRI ANNV (*.1 ARR. 'tube:150y of Illinois

Califotisia State San loseis," %Il, sioo i t, I Isis mil) of ColosasInpisril hi Viol. queens Collegeto.Nstn1 11. Stale iiscs+11)

DirsinAN Ilniteisity of IowaKEN:NE111h. SIT Nuffillon CaliforniaI II PM 101 W. SIMirvS; 'ilitc siltyTU05115 O. SI OAS, Ilniseisis) of Califoinialinloelcy

a a *.1

Speech Monographs is published in the months of March, June, August,and November by the Speech Communication Association, Statler Iltltan!bad, New York City 10001. Second-class postage paid at New York, NewYork and other mailing offict s.

Correspomknce concerning issues and permission should be addressedto the New York office. Correspondence concerning manuscripts should beaddressed to the Editor.

Copyright 1973 by the Speech Communication Association.Annual subscription, $20.00 (includes membership in the Speech Convo-

nmeocation Association). Single copies,443*. .5obIndexed and/or abstracted in Psychological Abstraru, Af Aboradi

Education Indrx, Language and Leittuage Behavior Abstracts, and Kxcetional Child Education Abstracts. ,PERMISSION To REPRODUCE THIS copY

RIGHTED MATERIAL HAS BEEN GRANTED 9Y

Speech CommunicatiorAasociatiOn

TO EPIC AND OFTGANI2A TIONS OPERATINGUNDER AGREEMENTS WITH THE NATIONAL IN-STITUTE OF EDUCATION. FURTHER REPRO-DUCTION OUTSIDE THE ERIC SYSTEM RE-QUIRES PERMISSION OF THE COPYRIGHTOWNER.-_

r SO

SPEECH COMMUNICATION

ASSOCIATIONFounded 1914

incorporated 1950:

OFFICERS AND STAFF

SAMUEL L. BECKER, President

HERMAN COHEN, First Vice-PresidentLLOYD F. Brrzins, Second Vice-President

WILLIAM WORK, Executive Secretary

ROBERT N. HALL, Associate Executive Secretary

University of IowaPennsylvania State University

University of WisconsinNational, OfficeNational Office

PATRICK C. KENNICOTT, Associate Executive Secretaryfor Research.

BARBARA LIES-BRILHART, Associate Executive Secretaryfor Education

EDITORS

PATRICK C. KENNICOTT, Editor,Bibliographic Annual

ROBERT L. SCOTT, Editor,The Quarterly Journal of Speech

THOMAS M. SCHEIDEL, Editor,Speech. Monographs

MARY M. ROBERTS, Editor,The Speech. Teacher

ALTON BARBOUR, Editor,Free Speech. Yearbook

DENNIS LYNCH, EditorNon-Print Media

WILLIAM WORK, Editor,Spectra

National Office

National Office

National Office

University of Minnesota

University of Wisconsin

Kansas State College of Pittsburg

University of Denver

Memphis State University

National Office

1974 ADMINISTRATIVE COMMITTEE

Samuel L. Becker, PresidentHerman Cohen, AM Vice- PresidentLloyd F. Bitzer, Second Vice-PresidentR. R. Allen Chairman,

Educational Policies BoardThomas Sloan, Chairman,

Publications BoardHerbert Simons, Chairman,

Research BoardJohn E. Dietrich, Chairman,

Finance BoardAnita Taylor, Member,

Finance BoardKenneth Andersen, Member,

Finance Board

University of IowaPennsylvania State University

University of Wisconsin

University of Wisconsin

University of California

Temple University

Michigan State University

Florissant Valley Community College

University of Illinois

SPEECHMONOGRAPHS

SPECIAL ISSUE:

Communication and Conflict

EDITORIAL BOARD FORTHE SPECIAL ISSUE

Editor

JOHN WAITE BOWERS

Department of Speech and Dramatic ArtUniversity of Iowa

Iowa City, Iowa 52242

Editorial Assistant

CONSTANCE SWANK

University of Iowa

Associate Editors:

Fred E. Jandt, State University College at Brockport, New York

Patrick C. Kennicott, Speech Communication Association

Gerald R. Miller, Michigan State University

C. David Mortensen, University of Wisconsin, Madison

Herbert W. Simons, Temple University

Thomas M. Steinfatt, Queens College of The City Universityof New York

Philip K. Tompkins, State University of New York at Albany

Speech MonographsPublished by THE SPEECH COMMUNICATION ASSOCIATION

VOLUME 41 MARCH 1974 NUMBER 1

Editor's Introduction:Beyond Threats and Promises

An Analytic Model of Conflict

JOHN WAITE BOWERS IX

CHARLES E. WATKINS 1

One of the major criticisms of conflict models has been thatthey fail to include such communication terms as threats,promises, credibility, and messages. This paper details an ana-lytic model of the conflict situation which defines these termswith action alternatives, value and perception functions, andtermination expressions. The utility of the model is demon-strated by the inclusion of new concepts, the description of afictional conflict, and the deduction of new relationships amonga set of axioms.

Conciliation and Verbal Responsesas Functions of Orientation and Threatin Group Interaction THEODORE JON MARR 6

Levels of orientation and threat behavior were manipulated incontexts where subjects thought that they were in consensus-seeking discussions. Dependent variables were subjects' con-ciliatory behavior and levels of orientation and threat exhibitedby the statements they chose. The high orientation conditionwas found to produce high orientation choices. The effect ofthreat level was complex. Women, in general, behaved morerationally than men to gain rewards.

The Effects of Substantive and Affective Conflictin Problem-Solving Groups MAE ARNOLD BELL 19

This study tests the consequences of verbally-expressed affectiveand substantive conflict in eliciting three characteristics ofstatementsaffective, substantive, and metadiscussional con-tent. It also tests trends across discussion segments as theyaffect those three characteristics of statements.

Communication in GameSimulated Conflicts:Two Experiments

THOMAS M. STEINFATT,

DAVID R. SEIBOLD, and

JERRY IC FRYE 24

The results of two experiments on communication in game sim-ulated situations are reported. The first experiment investi-gated cooperative behavior in a Prisoner's Dilemma game withfull communication allowed, and found that real rewards pro-duced significantly greater cooperation than did imaginary re-wards. A Creative Alternative game was used in the second ex-periment to examine the effect of communication in a situa-tion in which one of two parties has a strong motivation to getthe other person to change his behavior, but the other haslittle or no reason for doing so. The results suggested that boththe opportunity to communicate and the dogmatism of theparties may be related to the ability to achieve a creative solu-tion.

Argument in Negotiation: A Theoretical NANCY A. REICHFS

and Empirical Approach and HARRIET B. HPRRAL 36

Five argumentative dimensions of negotiationpower, risk,compromise, prediction, and situationare analyzed. Two ex-ploratory experiments testing the effects of prediction and ofone situational element, urgency as a function of time allotted,are reported.

Perceiving Communication Conflict THOMAS J. SAINE 49

Two experiments were undertaken to test the independent andinteractive effect of cognitive complexity and information loadon the perception of communication conflict and ratings of con-flict severity. Results provide support for hypotheses derivedfrom information processing theory which predict that highcomplexity decoders, under optimal conditions of informationload, are better able to perceive conflict and record higher esti-mates of conflict severity than low complexity decoders underthese same communication conditions. Information processingtheory was deemed an appropriate theoretical point of depar-ture for assembling a theory of conflict perception.

A Literary Analog to ConflictTheories: The Potential forTheory Construction

LAWRENCE J. CHASE

and CHARLES W. KNEUPPER 57

A case study of Dostoevsky's Crime and Punishment (Part I)was conducted to demonstrate the existence of literary analogs.to contemporary conflict theories, as well as to indicate the po-tential for theory construction via literary analysis. Three ana-logs were described which correspond to tht_Jrustration-aggres-sion model, the decision-making or game-th?oretic approach,and the instinctual aggression paradigm, respectively.

Communication and the Inducement ofCooperative Behavior in Conflicts:A Critical Review DAVID W. JOHNSON 64

Studies of conflict are critically reviewed for their treatment ofcommunication. The implications of these studies for com-munication theory are explicated.

SPECIAL REPORTS

Opportunity to Communicate and Social Orientationin Imaginary-Reward Bargaining JAMES G. GREENWOOD 79

Subjects at varying levels of opportunity to communicate(highly restricted, moderately restricted, and unrestricted) andsocial orientation (cooperative and competitive) played a gamewhose best outcome required an agreement to redistribute pay-offs. Main effects were found for both variables, with the pat-tern of means indicating that successful bargaining is mostlikely in a cooperative social climate permitting unrestrictedcommunication.

An Experimental Verificationof Schelling's TacitCommunication Hypothesis

THOMAS E. HARRIS

and ROBERT M. SMITH 82

Rationale and results are reported for a test of tacit communication among large university and small college students in-volving certain of Thomas Schelling's exercises. Schelling's pre-dictions are substantially confirmed among both groups regard-less of whether partners are "real" or "hypothetical" and ofprior knowledge of partner in the "real partner" condition.

AIM... 11 Mb. A.1.i*".4.111=11.0111,...11.101,

POLICY STATEMENT

Speech Monographs is a journal which publishes learnedinquiries concerned with speech and related communica-tion behaviors. Speech Monographs welcomes monograph-length papers for which it is the Speech CommunicationAssociation's only outlet, while at the same time consider-ing shorter reports, both theoretical and critical, and briefresearch notes. No speech communication related topic andno mode of inquiry is barred from consideration by SpeechMonographs.

Papers are evaluated for possible publication on the basisof whether the work reported in them substantially ad-vances understanding of topics associated with speech orrelated communication behaviors.

Manuscripts submitted to Speech Monographs should besubmitted in three copies, one of which should be a ribboncopy. Submitting fewer copies will delay consideration; amanuscript will not be considered without a ribbon copy.An abstract by the autlfor should also be included. The,author's name should apivir on a separate page so that theauthor's identity will floc be known to the Associate Editorswho will read the manuscript and make recommendationsregarding it.

The MLA Style Sheet, Second Edition, should be fol-lowed in the preparation of manuscripts. For resolution ofproblems not covered there, authors should follow therecommendations of A Manual of Style, Twelfth Edition,published by the University of Chicago Press. Authorsshould consult recent issues of Speech Monographs for ex-amples of appropriate style.

Authors will receive forms for requesting reprints. Or-ders must be placed with the Standard Printing Company,201 North Third Street, Hannibal, Missouri 63401. Re-prints cannot be obtained at a later time.

Manuscripts should be submitted to:

ROGER E. NEBERGALLEditor-Elect, Speech MonographsDepartment of Speech CommunicationUniversity of Illinois at Urbana-Champaign244 Lincoln HallUrbana, Illinois 61801

GUEST EDITOR'S INTRODUCTION:BEYOND THREATS AND PROMISES

I N May 1973 I was invited to delivera lecture at Bowling Green State

University in Ohio. That was a yearafter the SCA Research Board had spon-sored a conference for ten participantson The Role of Communication in theProcess of Conflict." The book resultingfrom manuscripts discussed at the con-ference was in preparation. (Its publica-tion by Prentice-Hall under the editor-ship of Herbert Simons and Gerald Mil-ler is now impending.) The researchBoard had sponsored, and I had organ-ized and chaired, a special competitiveprogram on communication and conflictat the annual SCA convention the previ-ous December. This special issue ofSpeech Monographs had been generous-ly authorized by editor Thomas Scheideland had just been announced. I and afew others were offering courses forgraduate students on communicationand conflict. All these activities were de-signed to stimulate interest among schol-ars, especially young scholars, in whatsome of us perceived to be a neglectedbut important area for the kind of re-search and theory at which we should beexpert.

Before I prepared my Bowling Greenlecture, I decided on a title"IleyondThreats and Promises"because I hadnoted that theorists of conflict in gener-al seldom attend to communicative be-haviors other than messages that clearlyfit into those two categories. (To givecredit where it ought to be given, I mustnote that James Tedeschi writes also of"mendations" and "warnings.") Havingmade up a title, I could begin my prepa-ration, and I commenced to scan my en-vironment for clues.

I decided to begin from the definitionsof threat and promise that are most

widely accepted among conflict theorists.Imagine two parties to conflict, Archerand Target. A threat exists when one(say, Archer) predicts that he will imposenegative sanctions on the other, thesesanctions to be contingent on some be-havior of the other. A promise simplychariges the sign of the sanction: Archerpredicts that he will deliver positivesanctions to Target contingent on somebehavior of Target's. A threat or apromise, then, to be credible and unam-biguously identifiable, would require:

1. that Target's behavior is somehow relevantto Archer's value position, and Target knowsit;

2. that Archer, at least under certain drcum-stances, his control of negative or positivesanctions for Target, and Target knows it;

3. that Archer is able to impose those sanctions,and Target knows it;

4. that Archer Is willing to impose those sanc-tions, and Target knows it;

5. that Archer sends a message to Target in-dicating Archer's intent to impose thosesanctions contingent on Target's behavior.

(This list of characteristics is less com-plicated than John Searle's analysis ofpromises, but it obviously owes some-thing to him.)

What did my environment tell me?My first observation was that both par-ties to conflict in everyday life often con-trol both kinds of sanctions for each oth-er and that the behavior of parties inconflict is best explained by their takingaccount of both rewarding and punish-ing contingencies. But English (for one)has no word to characterize a messageexpressing such a double contingency.So I invented one: "thromise."

My second, and most interesting, ob-servation was that adults in this culturerarely promise explicitly and even morerarely threaten explicitly. In the two or

three weeks of my observation, I neitherwitnessed nor was party to a single mes-sage formally marked as a prediction ofrewarding or punishing sanctions underthe control of an Archer to be deliveredcontingent on the behavior of a Target.(It should be noted well that we haveno very young children in our home.)Definitional condition (5) was nevermet. It is as though we have agreed notto say what we mean. Yet, during theperiod of observation I clearly witnessed(and was party to) situations where Tar-gets behaved as if threats, promises, andthromises had been delivered to them byArchers. Somehow, we frequently under-stand threats, promises and thromiseseven though, outside legal and religiousstructures, we seldom say them or hearthem.

How? I speculated at Bowling Greenthat Target will perceive a threat,promise, or thromise if definitional con-ditions (1.4) are met. I said then thatcondition (5) is redundant. I have sincechanged my position slightly. I now thinkthat some message is necessary (or, atleast, customary), but that if conditions(l-4) are met almost any message, includ-ing a grunt, a groan, or a grimace, willserve. We do not speak metaphorically,I think, when we use terms like "threat-ening person" and "promising situa-tion." The threat, the promise, or thethromise is in the interaction of Archer/Target's situations and Archer/Target'spersons. If Archer sends no message ex-ploiting that interaction, Target willsupply one. Among adults in this cul-ture, communication of threats andpromises is in general tacit, to employThomas Schelling's term, enthymematic,to employ Aristotle's. The situation andpersons of Archer and Target providethe premises for the pragmatic enthy-meme. Target has no problem drawingthe conclusion. When we do send threat-ening, promising, or thromising mes-

sages, their form is likely to be anythingbut a prediction. We equivocate ( "1

might like it if you . . ."), ambiguate("Isn't there something you can do. . ?"), and disclaim responsibility forconsequences ( "1 don't see how I coulddo anything but leave home if you ...").Form does not, match function.

How do we get that way? Our smallchildren are not at all reluctant to meetcondition (5), to make explicit threats,promises, and thromises. I think the an-swer is that we learn in this culture tovalue our interpersonal independence(or, for us cynics, our illusion of inter-personal independence) above almost allelse. Erving Coffman would say that ithas something to do with our definitionof "character," To act on the basis offormally expressed threats, promises,and thromises is to be controlled byothers, and we claim not to be con-trolled by others. We make some kindsof threats (extortion, blackmail) andsome kinds of promises (attemptedbribery) illegal, and we freely apply thelabels for those illegal acts to otherformal threats, promises, and thromises.To act on the basis of tacit threats,promises, and thromises, on the otherhand, is to be free, and we like to b,able to claim that we are free even itmaintaining that claim requires us toconform to the unexpressed (and pos-sibly misperceived) wishes of powerfulothers.

In such a system the formal expres-sion of threats, promises, and thromisesbecomes not only unnecessary but dis-functional. An explicit threat, promise,or thromise works against its author'smanifest intention. We learn not to beTargets and not to take aim. Our vil-lains threaten and promise. Our heroes'defiant actions speak louder than theirwords.

A number of interesting research ques-

tions arise from this analysis. Amongthem:

1. is the expression of threats, promises, andthromises more common interpersonally incultures whose theory is authoritarian andhierarchical than in cultures whose theoryis democratic and egalitarian? The analysisimplies that the answer Is "yes."

2. is the expression of threats, promises, andthromises a curvilinear function of inter-personal power? The analysis suggests thatthose who are very powerful have no needfor formal threats, promises, and thromises,and that those who are powerless cannotmake credible threats, promises, and thro-mises. The prediction is that the most com-mon such expressiOns emanate_ from thosewho have power in a middle range.

3. Does the expression and perception ofthreats, promises and thromises vary withpersonality types? Dogmatism (see Steinfatt,Seibold and Frye's paper in this issue),rnachiavellianism, and externality-Internalitymight be promising types to pursue.

4. Do predicted punishments (threats) and pre-dicted rewards (promises) behave additivelywhen they are combined in thromises? Ifthey behave nonadditively, are personalitytypes relevant to the way they function oneach other? The analysis makes no predic-tion about the answer to this question, butthe question is Interesting anyway. Is therea personality type for whom the salience ofreward (opportunity) is more potent thanthe salience of cost (risk) in a throtnisingsituation, and is there an obverse personalitytype? If such personality types exist, whatare the implications for communicative be-havior? A doctoral student at Iowa, Catty,trine Konsky, is pursuing such questions Inher dissertation. The personality variablespecifically of interest to her is success seek-ing/failure avoiding, together with difficultyof topic and power position, as predictors ofverbal behavior.

Now, about this special issue. I hadhoped that it would be filled with pa-pers ingeniously manipulating and ;Ana-lyzing the communicative behavior ofparties to conflict. It is not. David John-son's analysis of the communicativeshortcomings in conflict theory is nearlyas accurate after this issue as it was be-

fore. The most common communicationvariable treated by the authors repre-sented here is, as it has been among con-flict scholars for more than two decades,simple opportunity to communicate,rather than form and function of specificcommunicative behavior. But the situa-tion is promising. Most authors explicatecommunicative implications of their re-sults, and one might hope that they willsoon test those implications. A few stud-ies more explicitly testing hypothesesabout communication and conflict havebeen done but are not yet ready forpublication. Mae Arnold Bell, in a studybriefly reported, has extended TheodoreMares method so that conclusions canbe drawn about the production, not sim-ply the choice, of messages by subjects inpotentially conflictful situations. Her de-pendent variables are message variables.Gerald Miller and Thomas Steinfattnote the potential of Steinfatt's CreativeAlternative game for the generation andtesting of hypotheses about messages.Knowing the research habits of thosetwo scholars, I do not doubt that thepotential will expeditiously be madekinetic. Studies of communication inconflict will soon become so common, I.predict, that future special issues on thesubject would be gratuitous.

I want to acknowledge the contribu-tion of two individuals not mentionedelsewhere in the issue. Samuel L. Beckeredited one manuscript. But he owed mea favor. Eleanore Bowers, with no throm-ises explicit or tacit, displayed an unex-plainable but not unprecedented toler-ance for vagaries in the editor's tempera-ment and work habits occasioned by thisenterprise and others. I have alreadythanked the authors privately for put-ting up with my free editing and author-itarian ways specific to a crisis situation.I now thank them publicly.

--JOHN WAITE BOWERS

VoLumt 41.. _ 41.1.11.116.

MARCH 1974I& /Mk a.r. /IOW -.I* a Oh 410.1.

NUMBER i

AN ANALYTIC MODEL OF CONFLICT

CHARLES E. WATKINS

ONCEPTS find expression onlythrough patterns of symbols; thus

models provide researchers with handleson reality. But most conflict analystswould insist that a useful model is morethan a conceptual framework for or-ganizing physical phenomena: it is alsoa theoretical formulation for postulatingrelationships and predicting observableevents. And for quantitative purposes;it is also desirable for the model to util-ize a closed symbol system which pre-serves the important relationshipsamong the referents. Such models aretermed analytic and are commonplacein the physical sciences. For example,the formula PV = kT is an analyticmodel for certain properties of an idealgas, so that if P represents pressure, Vvolume, T temperature, and k a con-stant, then an increase in pressure (P)must result in an increase in tempera-ture (T) or a decrease in volume (V).The formula's algebraic properties cor-respond to the physical properties ofthe gas.

This monograph outlines an analyticmodel of conflict with a special emphasison the role of communication. Becausethe purpose of a model is to simplify

Mr. Watkins is a doctoral student in the De-partment of Speech Communication at theUniversity., of Texas at Austin and is presentlyseeking post-doctoral employment.

SPEECH MONOGRAPHS, Vol. f1, March 1971

complex phenomena, it is inevitablethat certain elements of conflict areomitted. Every effort has been made toinclude those closely related to tom-munication.1 Most of the essential con-ditions of conflict as reported in the lit-erature have been included, as sum-marized in the following axiomsts

1. Conflict requires at least two parties capableof invoking sanctions on each other.

2. Conflicts arise due to the existence of amutually desired but mutually unobtain-able objective.

3. Each party in conflict has four possible typesof action alternatives:a. to obtain the mutually desired objective,b. to end the conflict,c. to invoke sanctions against the opponent,d. to communicate something to the op-

ponent.4. Parties in conflict may have different value

or perceptual systems.5. Each party has resources which may be

increased or diminished by implementationof action alternatives.

1 Although communication takes place onmany levels and in countless settings, one of themost interesting contexts of the communicationact is In a conflict situation. Regardless of wheth-er the conflict is interpersonal or international,certain terms and elements remain important.Among these are threats, policies, messages,credibility, and reliability. Certainly there aremore, but this paper claims only to describe amodel accounting for those most often present.

2 The axioms have been selected from theessential properties of conflict advanoed by Ray-mond W. Mack and Richard C. Snyder, "TheAnalysis of Social Conflicttoward an Overviewand Synthesis," Journal of Conflict Resolution,I (1957), 212-248.

2 SPEECH MONOGRAPHS

6. Conflict terminates only when each party Issatisfied that it has 'won' or 'lost' or be-lieves that the probable costs of continuingthe conflict outweigh the probable costs ofending the conflict.

An implicit assumption is that all par-ties behave rationally, because it is onlyto the extent that actions are consistentand predictable that any model is ofusefor to speak of a rational model ispleonasm; to speak of an irrational mod-el is oxymoron.

THE SYMBOLIC REPRESENTATION OF THE

ELEMENTS OF CONFLICT

The first axiom postulates the ex-istence of two parties with potentialsanctions against each other. Althoughthe model could be modified to admitthe existence of more than two partiesto the conflict, that ,extension. wouldcompound the complexity of notation,withqut correspondingly enriching thetheory. The sanctions available, to agiven party are represented by the set

S12, , Stn),

where the superscript distinguishes, be-tween individual sanctions in the ar-senal. Rewards are the same as sanctionsexcept that they have positive value tothe recipient.

The second axiom stipulates the existence of an object for the conflict,something each party desires, but whichboth cannot have. This does not entailthat the object is the same for both par-ties, but rather that both parties cannotsimultaneously realize their objectives.Nor does this imply that conflict willcontinue only until one party can obtainthe objective for himself, because goalsmay be redefined as the conflict evolves.Nor may it be presumed that either ofthe conflicting parties will obtain theobject, for sometimes both sides agreeto cease hostilities before either can at-

tain his objective. The action of partyachieving his objective is represented bythe symbol 01, and employing the math.ematical convention of appending the(1) marker to negate the meaning of asymbol, 01 stands for the failure ofparty I to achieve his objective.

Third, the action alternatives open toparty i are represented by the set (0i,01, T, , C1(01, where Oirepresents the attainment of the mutual-ly desired objective, T represents thetermination of the conflict, (Stir . . ,

Sib) stands for the group of sanctionsavailable to party i, and Cirej representsthe communication by party t of. an ex-pression e, which may be a statement offatt,' intent, or condition.

The fourth axiom declares that theconflicting parties may have differingvalue or perceptual systems. This may beintroduced into the model by means oftwo functions Vi(X) and Pi(X). Vi(X)-represents the value of X to party i insome unit of value, so that values possessall of the mathematical properties of thereal' number system, such as addition,subtraction, and ordet. P5(X) representsthe perception of X by party i, giving aperception the ,same role as the objectperceived, so that a perteived value hasthe properties of a value, a perceivedaction has the properties of an action,and so forth. The model makes use of athird function, so it is probably wise tointroduce it here. This is the wellknownprobability function p. Thus, p(X) rep-resents the likelihood of X by a realnumber between 0 and 1. It is possiblefor X to be an action, a perception, avalue, or a conditional of the form A -9B, meaning "A will result in B" or "if A,then B." This makes it possible to repre-sent contingent policies by statementssuch as SI Si, which states that sanc-tion Si will be met by sanction Si.

The fifth axiom states that the partieshave resources which may be affected by

AN ANALYTIC MODEL or CONFLICT 3

the actions of either antagonist. Theamount of resources available to party Iis represented by ri and is expressed inthe same units as the value function forthat party. Gain or loss of the objectives,termination or continuation of the con-flict, or the deployment of a sanction orreward may effect the resources of bothparties, so that the value of a sanctionis the change in resources it produces.For example, a sanction which does greatdamage to an opponent's resources mayrequire the expenditure of the resourcesof the sanctioner in order to manufac-ture and deliver the weapon. Note alsothat the continuation of the conflictmay be of benefit to one or perhaps bothparties. Communications do not haveintrinsic value, but may cause benefits ordetriments indirectly by eliciting actionreplies. To be threatened does not en-tail direct cost, but to honor the threator to prepare defenses Might.

So far the actions available to bothsides have been defined, certain opera-tions on the actions have been expressedas functions, and a notational schemefor representing each has been described.The final axiom stipulates the limitswithin which these variables may bemanipulated by the definition of threeexpressions which determine the condi-tion for termination of the conflict. Thefirst two are intuitively obviousa partywill cease to fight when it has obtainedall that it can from the conflict or whenit concedes that there is nothing thatcan be done to improve its position.These conditions are dynamic in thatboth parties may continually revise their'win' and 'lose' expressions, W and L,but at any moment both will be definedand available to at least that party forcomparison with the situation at thattime.

The third condition for terminatingthe conflict is really a generalization ofthe first two, the belief that the proba-

ble costs of continuing the conflict out-weigh the probable costs of terminatingthe conflict. The grounds for this 'quit'condition originate in the central hy-pothesis of decision theorysthat con-flicting parties will strive to maximizetheir expected utility. When confrontedwith the decision to quit or to pursueone of the action alternatives, a partycompares the expected utility of termi-nation (p(T)V(T)) with that of the al-ternative (p(T)V(T'}], and chooses thelarger figure. The problem of evaluatingV(T) is solved by use of the other termi-nal conditions, W and L, in place of T.This gives an expression of the formp(W)V(W) + p(L)V(L) + p(T')V(T')dependent in value on the estimatedprobabilities and values of victory, de-feat, and the various strategic options.

Thus the axioms may be rewrittensymbolically:

I. There exist at least two parties such that foreach party I, there exists the set [S11, Sig,

2. There exist 01 and Os such that Os Oso

and 0 .4 0I

3. For each party 1, there exists [Os, T.St1 . Sib, Sten.

4. For each party 1, there exists Vs such thatV1(X) is a real number, and Ps such thatP1(X) has the same axiomatic- properties asX.

5. There exists for each party 1, an rs such thatif Vs(X)50. then X increases rs.

6. T if and only if for each party I, Wt, Li, orQt is satisfied.

r*"

An example illustrates the use of thissymbol set:

At the big poker game in Lilly's GruesomeGulch Saloon, the Two Gun Kid [party k] andBlack Bart (party b] sit dowITAR play for stakesof $50 in antes and the pleasve of a drink withMiss Lilly. Bart has an eye for Lilly (Vb(Ob) =$100J, who slips $20 out of his money pouch

a For a fine resume of the worts of the de-cision theorists see Ward Edwards, "BehavioralDecision Theory," Annual Review of Psychol-ogy, 12 (Palo Alto: Annual Reviews, Inc., 1961),473-498.

4 SPEECH MONOGRAPHS

every time he bets (Vb(r) = $20], but theKid is pure In heart (Vk(Ok) = $501. Each hasthe options of grabbing the stakes [Os], betting(TSS, folding coo, various 'raises Sib]or palavering (Cie)]. Bart starts with $500 cashatter his $25 dollar ante (rb = $500]; the Kidhas $250 in gold dust and his pappy's false teethworth another $50 (rk = $300]. The Kid needs$500 to pay off the mortgage on his ranch (Wk:rk = $500), but won't play It he loses his pappy'steeth (Lk: rk teeth']. Bart wants Lilly andthe antes plus the Kid's gold (Wb': rb = ¢800 4.Lilly), but couldn't care less about the teeth(Vb(teeth) $0]. Bart will quit the game beforelosing his grubstake (Lb: rb = $200].

After the draw, Two Gun grins as he holdsaces and eights (Pk(p(Wk) = .60]. Bart drawls,"Kid, I gotcha [Cbtp(Wd = in." but thinks tohimself that his three jacks might not be goodenough (Pb(p(Wb) = .75j). The Kid stays in (Qk:(.60) ($25a) (.40)($25a) = $5]. So does BlackBart (Qb: (.75) ($75 = $25a + Lilly) + (25)($25a) = $50].

COMMUNICATION WITHIN THE

CONTEXT OF THE ANALYTIC MODEL

The power of a model is reflected inits ability to account for phenomena out-side its original domain, and the test ofan analytic model is its capacity to repre-sent new concepts and relationships withthe original symbol set and to deducenew relationships among the originalconcepts. In this final section, the utilityof the present model is described andillustrated in both of these respects.

The scholar interested in conflict isprobably concerned with several of theseterms:4 message, threat, promise, hones-ty, reliability, credibility. Each of thesecan be conveniently represented withinthe framework advanced in the previoussection. A message is a communicationand is easily expressed as the term e inthe communication action defined inaxiom (3), C[e]. A message which assertsthat party i has available a sanction

4 In fact, the major criticism of the decisionmodels has been their inability to account forthese communication terms. Schelling and Rapo-port in particular have solicited accounts forthreats, promises, and commitments.

which will diminish the resources ofparty j by one hundred units would beV1(St) 100 and the act of communi-cating that message would be C1[Vi(S1))= 100. A threat is the communication ofa conditional sanctioni.e. S1 will resultin the event of some condition, say theacquisition of the objective by party j:Ci[Oi -0 SI]. Implicit in the notion of athreat is at least a potential harm to thethreatened under certain conditions, sothat the value of the sanction, V1(S1), isnegative. Promises work the same way asthreats, except the sanction has an ilia-plicit positive value, so that if reward Siwere available to party i, he might offer .-

in return for being allowed to attainhis objective: C1[01 Si]. A party's reli-ability is the likelihood that what he issaying is true, expressed by p(Ci(X]X). His honesty is the probability that hebelieves what he says: p(C1[Xj --) Pi(X)).His credibility, is his reliability as per-ceived by his opponent: P1(p(CAX3X)). This portrays credibility in terms ofa statement, its, likelihood of being true,and the way it is understood by theother party. Thus several familiar com-munication terms can be isolated withinthe framework of the model.

The new terms fit in turn into themodel. A party to conflict may calculatewhether to honor a threat by consider-ing the potential damage if the threat iscarried out and the likelihood that theopponent will actually do as he says.This yields an expression of the form:[V1(S1)][P,(p(C1[X St] Si)))].This makes it clear that the threatenerwill seek to alter his opponent's percep-tions about his own reliability, perhapsby additional messages aimed at con-vincing his opponent that he meanswhat he says or by invoking otherthreatened sanctions or fulfilling prom-ises to demonstrate a high credibility. Orhe may seek to enhance the effectiveness

AN ANALYTIC MODEL OF CONFLICT 5

of his threat by persuading his opponentthat the damage threatened would be anintolerable loss against the potentialgain. But there are also costs in carryingout a threatened sanction, so figuring inVi(Si): [VI(Si) + Vi(Si)][1)1(p(C[X -. Si] -0(X --) Si)))]. Now a counterthreat of theform Ci(Si S1] is seen to reduce the effec-tiveness of the threat by simultaneouslyincreasing Vi(Si) and lowering the prob-ability of the original threat by means ofthe deterrent value of the counterthreat.

Meanwhile in Gruesome Gulch, Two GunKid makes the first bet of $50 [Ski; V1,(Ski) =$501. Ban eyes the glittering gold and sneers,"I think yer bluffing (C.biPb(Ck(Wk) - Wk) =0)1)." and bets $100 (VI Vk(Sbi) = $100]. TheKid looks at his cards and raises another $50iSk2; Vu(Sk2) = $150]. Black Bart raises back$100 [Sb2; Vk(Sb2) = $150] and growls, "Raiseagin an' I'll double it agin (Cb[Sk3 SO;V1,(5b3) = 2Vb(Sk3)11, 'cause there's a thousandmore in my pouch [Cb(rb = $1000]." Lilly isnow $40 richer [Vb(P) = $40].

In general, if a party to conflict con-siders all possible actions by both him-self and his opponent which bear on hisresources, his Q-expression will become:

Qt: P(Wt)V1(W1) + po,"(11) + PT)Vi(V) + p(S1)V1(S1) + p(Si)Vi(Si). Thusthe process of conflict becomes themanipulation of these expressions byboth parties by means of the action al-ternatives available to each. Both sidesmust estimate the probable actions, per-

ceptions, and values of the opponent inorder to achieve victory,

Back at the poker game, the Kid's bankrollis shot irk = + teeth) and he thinks he isbeaten [Pk(p(SN'k) = .20)1. it he folds, he loses$175 (Qk: (0)($275) + (1)($175 = $25a $50b$100c) = $1751. and if he calls he's not muchbetter off (Qk: (.20)($275) + (.80)($275 =$25a$50b$100c$100d) = $165). Bart issitting easy (Qb: (.75)($525 = $253+$50b+$100c+$1004 + 1,114) + (25)($275 = $25a--$50b$100c$100d) + (1) ($40) = $135] until hehears the Kid say, "I'll see your hundred [Sk3;Vb(Sb3) = $100] and I'll raise you my pappy'steeth [SO: Vb(Sk4 = $50]."

Bart knows the Kid is a good judge of cards(1)b(Pk(Vk) -) Wk)] and that he holds his pap.py's teeth in high regard (Pb(Vk(teeth) = $50))so he figures the Kid for a winner [Pb(p(Wk) =.90)]. As he removes Lilly's hand from his moneypouch (Pb(Vb(T) = $20)], Black Bart grumbles,"What's the use? [Qb: (10)($275) + (.90)($325)+ (1)($60) = $325]. I'll fold [Qb: (0)($325) +(1)($275) + (1)( $40) = $315)." And so asthe game ended ['F], the Two Gun Kid savedhis ranch [Wk), Black Bart returned to the hills[Lb], and Miss Lilly drank alone.

By representing some of the most im-portant terms outside those which werebuilt into the model, and by demon-strating new relationships among its ele-ments, this analytic model has been dem-onstrated to be a useful and powerfulmethod for approaching the study ofconflict. This is not to say that it cannotbe refined and expanded for differentfoci, but indeed is to justify its furtheruse by analysts of conflict.

CONCILIATION AND VERBAL RESPONSES ASFUNCTIONS OF ORIENTATION ANDTHREAT IN GROUP INTERACTION

THEODORK JON MARR

T EWIS COSER, in formulating Georg1.41 Simmers classic, Conflict and TheWeb of Group-Affiliations, into specificpropositions, provided anchoring pointsfor much of the present research on so-cial conflict.1 Simmers central thesis isthat "conflict is a form of socialization"which "means . . . no group can be en-tirely harmonious, for it would then bedevoid of process and structure."2 Oneof the major themes developed by Coserand Simmel is in-group conflict andgroup structure stabilization.

Many researchers have investigated thevariables which lead to the stabilizingoutcome,, or the consequent consensualsituation:1 size of group,* socioeconomicbackgroufid, aesthetic preference,* andthe amount of talking by members ofthe group.*

Mr. Marr is visiting professor at the Nevi; AsiaCollege, Chinese University of Hong Kong, onleave from the University of Virginia.

I Lewis Coser, The Functions of Social Con-Met (Glencoe, Ill: Free Press, 1956); GeorgSinunel, Conflict, trans. Kurt H. Wolff and TheWeb of Crougo-Affillatiotu, trans. ReinhardBendbt (Glencoe, Free Press, 1955).

Coser, p. 31.3 A. Paul Hare, "A Study of Interaction and

Consensus In Different Sized Groups," Ameri-can Sociological Review, 17 (1962), 261-267.

Albert loan Lott, "Group Composition,Ccintiiiiiiikatfori, and Consensus: An Investiga-tion According to Newcomb's Theory of Com.munication," Diss. University of Colorado 1958.

5 Robert F. Bales, "The Equilibrium Problemin Small Groups," in Working Papers in theTheory of Action, by Talcott Parsons, RobertF. Bales, and Edward A. Shils (Glencoe, III.:Free Press, 1953), pp. 111-161; Bernard M. Bass,"An Analysis of the Leaderless Group Dis-cussion," Journal of Applied Psychology, 33(1949), 527-533; Bobble Norlieet, "InterpersonalRelations and Group Productivity," Journal 01Social Issues, 4, No. 2 (Spring 1948), 66-69;Henry W. Rlecken, "The Effect of Talkativeness

SPEECH MONOGRAPHS, Vol. 41, March 1974

Discussion often is used to resolvegroup conflict. Thomas and Fink dem-onstrated that there is a trend towardunanimity during problem-solving dis-cussion.* In this study I am particularlyconcerned with the relationship betweenmessage variables in a conflict-producingsituation and the consequent concilia-tory behavior exhibited' ofthe group working toward consensus.

A theoretical solution can be derivedfrom Thibaut and Kelley's psychology ofsocial groups./ They posit that bothgroup discussion and group problem-solving are cost and reward processes.Their basic proposition is that whentwo or more people interact, each electsto- ,behave in a way which will providehim with the greatest reward and leastcost. Any behavior is both potentiallyrewarding and costly.* Thibaut andKelley have also incorporated Heider'sattribution theory in this reward andcost paradigm to explain how prior in-

on Ability to Influence Group Solutions ofProblems,' Sociometry, 21 (1958), 1109-321:Richard L. Lucas and Cabot L. Jaffee, "Effectsof HighRate Talkers on Group Voting Behavlorin the Leaderless-Group ProbleMSoIving Situ.ation," Psychological Reports, 25 (19M), 471477; Cabot L. Jaffee and Richard L. Mien."Effect of Rates of Talking and Correctness onDecision on Leader Choke in SmallJournal of Social Psychology, '79 (1969), 247.254;Edwin J. Thomas and Clinton F. Fink, "Modelsof Group Problem Solving," Journal of Ab-normal and Social Psychology, 63 (1961), 53-63.

a Thomas and Fink, 53-63.I John W. Thibaut and Harold H. Kelley,

The Social Psychology of Groups (New York:Wiley, 1959); Harold H. Kelley and John W.Thibaut, "Group Problem Solving," In TheHandbook of Social Psychology, ed. GardnerLindley and Elliot Aronson, 2nd ed. (Reading,Mass.: Addison-Wesley, 1969), 1.101.

a Thibaut and Kelley, pp. 100-123.

CONCILIATION AND PSRBAL RESPONSES

formation is a determinant of the con-sequent behavior adopted to maximizereward.'. Using the prisoners' dilemmagame, they showed that information iscrucial to a person's success in adaptingto situations of outcome dependence. Anindividual is said to be informatlonallydependent on another if the other per-son can raise his stability of attribu-tion to a higherlevel than he can attainfrom alternative information sources.Furthermore, Morton Deutsch, using theprisoners' dilemma game, substantiatedthat opportunity to communicate in-creases the tendency to choose cooper-atively.12 --

Two key sets of variables seem to havebeen su:,:ested by these researchers: (1)group member variablesevaluative dis-positions, size, personality, cohesivenessof the group, and (2) communicationvariables.

ORIENTATION AND THREAT STATEMENTS

Previous research by Bales and Strodt-beck suggests that two types of state-ments are of particular importance ingroup communication, orientation state-ments and threat statements." Threatalso has been found by Deutsch andKrauss to be an important factor in in-terpersonal bargaining.12

9 Kelley and Thibaut, 10-13.10 Morton Deutsch, 'The Effect of Motiva-

tional Orientation Upon Trust and Suspicion,"Human Relations, 13 (1960), 123439.

11 Bales' Interaction Process Analysis was firstpublished in Robert F. Bales, interaction ProcessAnalysis: A Method for the Study of SmallGrauPt (Cambildge, Addison - Wesley,1950). For his later version and other relatedstudies see Robert F. Bales, "The EquilibriumProblem," in Personality and Interpersonal Be.havior, by Robert F. Bales (New York: Holt,1970). For an Interesting example of the useof Bales' IPA'to test group variables see Ray-mond A. Katiell, Charles E. Miller, Naomi G.Rotter, and Theodore G. Venet, "Effects ofLeadership and Other Inputs on Group Proc-esses and Outputs," Journal of Social Psychol-ogy, 80 (1970), 157.169.

12 Morton Deutsch and Robert M. Krauss,"The. Effect of Threat Upon InterpersonalBargaining," Journal of Abnormal and SocialPsychology, 61 (1960), 181-189.

In the present study, high orientationverbal behavior is defined as a verbalstatement which includes one of the fol-lowing: (1) a procedural suggestion witha relevant fact to resolve conflict, or (2)a new fact relevant to an earlier pro-cedural suggestion. Low orientation ver-bal behavior is a verbal statement whichdoes not include either (I) or (2).

1 have defined high threat verbal be-havior as a verbal statement which re-flects the unwillingness of the source toconsider other alternatives and whichincludes words of antagonism towardother individuals br toward their ideas.Lbw threat verbal behavior is a verbalstatement which does not express thesource's unwillingness to consider otheralternatives and which does not includewords of antagonism toward other in-dividuals or their ideas.12

Thibaut and Kelley's theoretical ra-tionale provides reasons to postulatethat an individual's conciliatory behav-ior is affected by the information varia-bles in a consensus-seeking discussion.From the Gouran, Knutson, and Klinestudies we learn that high orientationverbal behavior facilitates conciliatorybehavior in a consensus-seeking situa-tion." Studies by Leathers, Bales,Scheidel and Crowell, and Guetzkow and

59 In the study by. James C. McCroskey andDavid W. Wright, 'The of anInstrument for Measuring Interaction Behaviorin Small Groups," Speech Monographs, 38(1971), 333.340, the six factors are; orientation,tension, flexibility, relevance, Interest, and ver-bosity. Tension and flexibility combined con-note the idea of threat.

14 Dennis S. Gouran, Wadable: Related toConsensus in Group DiscOssion of Questions ofPolicy," Diss. University of Iowa 1968. A con.densation of this dissertation appeared inSpeech Monographs, 36 (1969), 387-391. ThomasJ. Knutson, "The Influence of Orientation Be-havior on Reaching Small Group Consensus,"paper presented at the annual Western SpeechAssociation Convention, Fresno, Calif., Novem-ber 22.24, 1971; John A. Kline, "Indices ofOpinionated and Orienting Statements In Prob-lem-Solving Discussion," Speech Monographs,37 (1970), 282.286, and "Orientation and GroupConsensus," Central States Speech Journal, 23(1972), 44-47.

8 SPEECH MONOGRAPHS

Gyr consistently show that tension andnegative emotions usually are followedby periods of confusion, disorientation,and member dissatisfaction.ns Deutschand Krauss found that threat reduces thelikelihood of two individuals cooperateing. In other words high threat verbalbehavior in a discussion should reducethe amount of conciliatory behavior.

On the basis of these previous experi-mental studies, three predictions weredeveloped to be tested in this study:

1. In a consensus-seeking context, there is aninteraction between orientation and threat.Orientation has a greater effect on concilia.tory behavior when there is low threat thanwhen there is high threat behavior in thegroup,

2. High orientation verbal behavior evokes agreater degree of conciliatory behavior thanlow orientation behavior in a consensusseeking context.

3. Low threat verbal behavior evokes a greaterdegree of condliatory behavior than highthreat verbal behavior in a consensus-seekingcontext.

From Thibaut and Kelley's theoreticalrationale it can be reasoned that an in-dividual will conform to those informa-tion behaviors which maximize the prob-ability of high reward. Because the com-bination of high orientation and lowthreat verbal behavior is predicted toevoke the greatest amount of concilia-tory behavior, regardless of the verbalbehavior of the other members of thegroup, an individual will tend to re-spond with high orientation and lowthreat statements.

11 Dale G. Leathers, "Process DisruptiotvandMeasurement in Small Group Communication,"Quarterly Journal of Speech, 55 (1969), 287.300;and "Testing for Determinant Interactions inthe Small Group Communication Process,"Speech Monographs, 38 (1971), 182.189; ThomasScheidel and Laura Crowell, "Feedback inSmall Group Communication," Quarterly four.nat of Speech, 52 (1966), 273.278; Harold Guetz.kow and John Gyr, "An Analysis of Conflictin Decision-Making Groups," Human Relations,7 (1954), 367-382. See also Bales, InteractionProcess Analysis.

On the other hand, Deutsch andKrauss found that If a person usesthreat. in an attempt to intimidate an.other, the threatened person . . . wouldfeel hostility ,toward the threatener andtend to respond with counterthreat and/or increased resistance to yielding."14On the surface, this seems to predictthe opposite of what Thibaut and Kelley's theory would. Consideration of thenature of threat renders the apparentdiscrepancy insignificant. Threat, as it isapplied in different situations, can be ofdifferent degrees. The relative strengthor weakness of a threat depends on thethreatened "Fertiin'sinternalf,tationatwell as on the particular threat elementused. Whether the person will respondwith counterthreat depends on whetherhigh threat also reduces conciliatorY be-havior. In other words, a predictionabout the individual's threat responsehas to be consistent with the predictionabout his conciliatory response. If pre-diction 8 is valid, the individual in ahigh threat condition will respond withgreater threat than the one in a lowthreat condition, Thus, predictions 4, 5,6 and '7 would follow:

4. The interaction between these two inde-pendent variables, orientation and threat,is such that the descending order of theireffects in evoking orienting behavior in aconsensusseeking context is high orientstion-low threat, low orientationlow threat,low orientation-high threat, high orienta-tion.Iow threat. There is no basis for pre-dicting a main effect of orientation con-ditions on orientation responses or threatresponses. Therefore, no prediction is madefor these variables.

5. If prediction 3 is tenable, the overall highthreat condition evokes less orienting be-havior from the naive subject than does theoverall low threat condition.

6. The interaction between these two inde-pendent variables, orientation and threat, issuch that the descending order of theireffects in evoking threat behavior In a con.

la Deutsch and Krauss, 182.

CONCILIATION AND VERBAL RESPONSES

TABLE I1REststro Seurnutat

LettersReward:

0 Y$1.55 WO $0.85 $0.60

sensus-teeking context is: high orientation-high threat, low orientation-high threat, loworientation-low ,threat, high orientation- lowthreaL

7. If prediction 3 is tenable, the overall highthreat condition evokes more threateningverbal behavior from the naive subject thandoes the overall low threat condition.

No prediction was made about the ef-fect of sex, but since it was controlleda test of its effect was made.

DESIGN AND PROCEDURE

The central idea was to construct anexperimental design which would sim-ulate a group discussion in which thediscussants had apparent opinions andwere trying to achieve consensus. Theywere given the illusion that they wereexchanging previously written notes totry to arrive at consensus and, in so do-ing, they anticipated being rewardedwith money. The monetary reward wasintended to be a manipulated analogueof the pressures which cause individualsto discuss and move toward consensus.Though it was introduced externally, itwas not external coercion.

The experimental sessions were ar-ranged so that four naive subjects,strangers to each other, arrived at thesame time. They were seated in front ofa semicircle, of partitioned booths, withthe experimenter at the center of thesemicircle. Each booth had a small slotin front of the subject and experimen-ter.st

The experiment was designed to con-

is Most of the sessions were arranged so thateither all the subjects were male or all werefemale. Due to . the difficulties in schedules,two experimental sessions had some of each.However, there was an equal number of malesand females in each of the experimental con-ditions.

B P$0.55 $0.10

trol the communication received by.eachof the four naive subjects at a patio,ular session. The particular set of corn-munication statements each receivedwas determined by the experimentalcondition to which he had been ran-domly assigned. The four experimentalconditions were: low orientation-lowthreat, low orientation-high threat, highorientationlow threat, high orientation-high- thrtat.is Eathgobjettvai led to be-lieve that he was sending messages tothe three other persons in the room withhim and that he was receiving messagesfrom those other persons, with the ex-perimenter as the intermediary in thecommunication channeling effort. Infact, each one of them was carrying ona "discussion" with three fictitious per-sons whose communication output wascontrolled by the experimenter and cor-responded to one of the four experimen-tal conditions.lo

As Table I shows, in the rewardschedule a monetary value was asso-ciated with each of the six letters.20 Allthe experimental conditions involvedfour "discussants," two of whom held toone extreme position white the othertwo held to the other extreme position.To simulate such a distribution, the

18 The four independent conditions wereestablished through pilot studies to ascertainthat these messages were in fact rated accord-in to their designations.

lg Anticipating that some student volunteersmight not show up for the experiment, a fewgraduate students served as standbys to takethe empty chair in order to simulate the four-person discussion group.

20 In order to maximize the difference be-tween consensus and nonconsensus, a sixpointreward scale was adopted to prevent the obviousmid-point compromise. It was hoped that thenaive subject would pick It initially when hewas asked. All those subjects who did not chooseR at the first choosing were discarded from theanalysis.

10 SPEECH MONOGRAPHS

TABLE 2PIACENZA REWARD SCIMULES root Z EXPERMENTAL St.SSION

R 0 . P

$0.10 $0.35 $0.60 $0.85 31.10 $1.35Si 0.10 0.35 0.60 0.85 1.10 1.35

1.35 1.10 0.85 0.60 0.35 0.10S4 (subject) 1.33 1.10 0.85 0.60 0.35 Q.10

TABLE STALLY &MIT or PLAYZIS' MUT CHOICE

SiStSsSA

R

ROOO.

YY

BBBB

Chotten letter.

basic reward schedules for the four "dis-cussants," one subject and three ficti-tious persons, were as shown in Table2.21 Subje'cts were also informed thatall reward schedules were not necessarilythe same.

The experimental Session consisted offour stages: pre-discussion letter choice,discussion, post- discussion letter choice,and final questionnaire. At the pre- andpost-discussion letter choice stages, sub-jects were asked to indicate the letterwhich they would choose. The pre-dis-cussion choice is only a preliminary in-dication of each one's preference. Afterthis initial preference indication eachWAS informed of the preferences of theother,"mernbers" by a tally sheet. (SeeTable 3 for the tally sheet used in thestudy.) Of course, the tally sheet re-corded each subject's response and theexperimenter's pre-determined responseof the three fictitious members.

The "discussion" consisted of twentystatements. Each subject was allowed toinitiate five of these (statements 3, 7, 10,18, 20). When it was his turn to initiatecommunication, he was given a set offour statements from which he had tochoose one to be sent to the other three

21 The amount of monetary reward requiredto arouse sufficient incentive was determined ina pilot study.

participants. The four subjects at eachsession were led to think that they wereexchanging notes with each other; infact, the notes which they received werepredetermined by the experimenter ac-cording to the orientation and threat ,

conditions which were Assigned random-

After they had "conununicatedriheymade their second choice. They weretold beforehand that if all ,four agrkedon their choice, each would receive award according to his reward schedule.If they did not agree, they were told,none would get anything. After they hadmade their second choice, each was givena questionnaire designed to find outwhether he suspected the true intent orprocedures of the game,

Each subject was allowed to choose,whenever it was his turn to communi-cate, among four prepared statements..These statements were -constructedsuch a way that they could fit equallywell into any of the four conditions. Thefour statements represent the four pos-sible combinations of the independentvariables: high orientation-low threat,low orientation-low threat, high orienta-tion-high threat, and low orientation--high threat. Thus, a subject could re-spond with high or low orientation, andhigh or low threat.

CONCILIATION

STATISTICAL DESIGN AND

DEPENDENT MEASURES

Essentially the experiment is a 2 x 2x2Independent groups design with two ley.els of orientation and two levels of threatand two sexes. The dependent concilia-tory measurement for predictions 1, 2,and 3 is taken from the subject's finalchoice. The dependent scores are: R =0, 0 = 1, Y = 2, G 3, = 4, and P=6. These are equivalent to the numberof spaces which a subject had movedfrom his first choosing, since only thosewho chose R Initially were included inthe analysis.-

AND VERBAL RESPONSES

RESULTS AND DISCUSSION

Subjects were 62 college students. For

the final analysis, however, there wereonly fourteen subjects in each cell, sevenmales and seven females, almost all ofwhom participated in singlesex groups.Tables 4 through 6 list the cell means,marginal means, and analysis of variancesummary table.

There was significant interaction be-tween orientation and threat. There wasrelatively more conciliatory behavior inthe high threat than in the low threatcondition when orientation was low.Prediction 1, therefore, was not con-firmed,

Suhjects receiving high orientationstatements exhibited sign can y greaer conciliatory behavior than subjects re-ceiving low orientation statements. Thehigh orientation subjects moved an av-

TABLE 4CELL MEANS OF CONCILIATORY BEHAVIOR

LowOrientation

HighOrientation Marginal

MaleLow ThreatHigh Threat

FemaleLow ThreatHigh Threat

n -7

1.71 2.290.43 2.29

2.141,43

2.713.29

1.68

2.39

TABLE 5ORIENTATION AND THREAT MARGINAL MEANS or CONCILIATORY BEHAVIOR

Low HighOrientation Orientation Marginal

Low Threat

MarginalThreat

D .2-- 14

1.930.931.43

2.302.792.64

2.211.86

TABLE 6ANALYSIS OF VARIANCE SUMMARY- TABLE or CONCILIATORY BEHAVIOR

Source cif Mean Square FRatioentation 0)

ThOri (-reat (T)Sex0 X

(3)

0 XS X TOXTXSWithin

1

1

1

48

20.641.797.145.790.01.140.00.76

26.472.29.9.167.42'0.01,470.0

Signifieant beyond the .05

12 SPEECH MONOGRAPHS

erage of 2.64 spaces, which is slightlymore than half of the total possible dis-tance (2.5 spaces). Prediction 2 was thusconfirmed.

Prediction 8 was not confirmed. Therewas no significant difference betweenthose who received high threat state-ments and those who received low threatstatements in their conciliatory behavior.

Female subjects exhibited significant-ly greater conciliatory behavior thanmale subjects did. The mean amount ofcompromise of the female subjects (2.89spaces) was almost as large as that of thetotal high orientationsp

group (2.64_aces).Of predictions 1, 2 and 8, only predic-

tion 2 was confirmed, This is consistentwith field study findings of Gouran,Knutson, and Kline, that high orienta-tion verbal behavior evokes a greaterdegree of conciliatory behavior than loworientation verbal behavior in groupsseeking consensus.

Though my theoretical rationale sug-gests, that high threat statements evokeless conciliatory behavior than lowthreat statements in a consensus-achieve.ing context (prediction 8), the resultsshow no significant difference. The basisfor prediction 8 was the finding fromprevious research that threat reduces co-operation 22 The unexpected outcomemay be due to different ways that anindividual can internalize the attributesof a discussion situation. The threatenedperson can react by offensive measures.The offensive tactic is not to yield to thedemands of the threatener, i.e., not toconciliate. At the same time, the threat-ened can counter with high threat verbalresponses. This is the alternative sped-

22 Deutsch and Krauss. 181-189. Also see J. L.Loomis, "Communication, the Development ofTrust and Cooperative Behavior," Human Re-lations, 12 (1959), 305.315; Alvin Scodcl, J.Sayer Minas, Phtlbum Ratoosh, and MiltonLipetz, "Some Descriptive Aspects of Two-Person Non-Zero-Sum Games," Journal of Con-flict Resolution, 3 (1959), 114-119.

fled in prediction 3 and its complemen-tary predictions 5 and 7, and would re-sult in no monetary reward. On theother hand, the threatened may use adefensive rather than an offensive strat-egy for coping with the situation. Thecoping or defensive measure is to give inbecause by conciliating one may at leastreceive a small monetary reward.

The determinant of which type of re-sponse will be selected in this contextseems to be the degree of threat whicha discussant perceives. In a situation per-ceived as extremely threatening, therecan be a boomerang effect and thus theoffensive tactic. Under tolerably threat-ening conditions, the reaction is to copewith the situation. My assumption isthat most of the subjects in my highthreat condition did not perceive thethreat as intolerable. This explanationcan be tested by replicating this studyincluding much higher levels of threat.I would expect a quadratic componentin the main effect of threat.

Although there is a significant inter-action between orientation and threat,it is not in the direction prediction 1indicated. The obtained significant in-teraction is due to an increase in thelevel of orientation, increasing concilia-tory behavior much more sharply whenthreat was high than when threat waslow. This interaction, though not sug-gested by Thibaut and Kelley's rewardand cost paradigm, can be explained bythe drive hypothesis.23 Threat is a sourceof drive. High threat statements evokegreater drive than low threat state-ments. High threat statements, therefore,increase the emission of dominant re-sponses by increasing the individual'slevel of general drive. The dominant re-sponse in this context, according to theThibaut and Kelley paradigm, is thatresponse which the individual expects

23 Robert B. Zajonc, "Social Facilitation,"Science, 149 (1965), 20-274.

CONCILIATION AND

to give him the greatest reward. In thehigh orientation condition the dominantresponse is compromise, while in thelow orientation condition compromiseis the subordinate response. Therefore,high threat statements increase concilia-tory behavior more as orientation is in-creased than low threat statements do.

To further probe the effect of the in-dependent conditions upon the subject?conciliatory behavior, I compared theestimated variances of the conciliatorybehavior in each of the conditions. Thefindings thus obtained were not pre-dieted by previous studies. Tables 7through 10 show the estimated variancesfor each of the conditions and the F-testsof the differences among them. Esti.mated variance is a measure of the ho-mogeneity of the behaviors of the mem-bers of a group after exposure to anexperimental treatment. The larger the

TABLE 7ESTIMATED VARIANCES OF ORIENTATION

AND THREAT LEVEES

Low HighOrientation Orientation

Low ThreatHigh Threat

1.08 0.250.64 1.311

estimated variance the greater is the di-versity of responses in a group. Com-parisons between pairs of groups, as in-dicated in Table 8, show that at the highorientation level, as we go from lowthreat to high threat, estimated varianceincreases significantly; whereas at thelow orientation level, as we go frbni thelow threat to high threat, estimatedvariance does not increase significantly.

VERBAL RESPONSES 13

(As a matter of fact, in this experiment,it decreased, but not significantly.) Thisseems to indicate that, as threat wasvaried in the low orientation conditions,subjects responded to each threat con.dition fairly uniformly. In the high ori-entation condition, the subjects did notrespond to the high threat statementsuniformly. In fact, the increase in theestimated variance indicates that somegave in to threat very much while oth-ers counteracted, bounced back andclung to their original positions.

The comparison in Table 9 indicatesthat the, male subjects were, responsiblefor the significant difference in esti-mated variances among the conditions.

TABLE 9ESTIMATED VARIANCES OF ORIENTATION AND

THREAT LEVELS SEPARATED BY SEX

LowOrientation

HighOrientation

Male SubjectsLow Threat 1.63 0.20High Threat 0.25 1.63

Female SubjectsLow Threat 0.41 0.20High Threat 0.53 0.49

Male subjects, when confronted with loworientation-low threat statements or highorientation-high threat statements, didnot respond uniformly in their concilia-tory behavior. On the other hand, malesin the low orientation-high threat con-dition in general stuck with their orig-inal extreme choice, while males in thehigh orientation-low threat conditionall tended tcrrnOve to roughly the samecompromise position toward the middleof the reward schedule. Almost all fe-

TABLE 8COMPARISON OF ESTIMATED VARIANCES OF ORIENTATION AND THREAT LEVELS

Low Orient.-Low rAsreat

High Orient. -High Threat

Low Orient.-High Threat

High Orient.-High ThreatLow Orient. -High ThreatHigh Orient: Low Threat

1 221.684.300

2.055.25' 2.56

'Significant beyond the .05 level.

- 14 SPEECH MONOGIMPHS

TABLE 10COMPARISON OF ESTIMATED VAIMANCEA OF MALE. SVDJECTS

sY ORIENTATION AND THREAT LEYELE

Low OrienteLow Threat

High Orient.High Threat

Low Orient.High Threat

High Orient; High ThreatLow Orient...High ThreatHigh Or lentLow Threat

'Significant at the .05 level.

1.006.670 6.708.000

soganSelzalatttar8,000 1.20

TABLE 11Cat. MEANS OF CELUENTATION RESPONSE

LowOrientation

hOrienHigtation Marginal

Male 3.18LoW ThreatHigh Threat

2.293.00

4.293.14

Female 3.68Low Threat 3.00 4.71High Threat 2.43 4.57

Ng 7

males, on the other hand, took roughlythe same compromise position towardthe middle of the reward schedule un-der all the conditions.

As one might infer from these results,females overall exhibited significantlygreater condliatory behavior than males.(See Table 6.) This suggests that maleswere less logical than the females, whounder any condition chose to com-promise, since all subjects knew thatthey would receive no monetary rewardunless there was consensus:

Tables 11 through 13 show that sub-jects who received high orientation statements averaged 4.18 high orientation re-sponses, while those who received loworientation statements only averaged2.68 high orientation responses. Thisresult is consistent with those found

by ,Knutson 44 In his experimentalgroups, the confederate introduced eith-er high, low, or no orientation. After theexperiment he asked the discussants torate each participant on his contributionto orienting the grouP'. Knutson foundthat in the high orientation conditionthe subjects not only rated the ccntederate high in orientation but also theirfellow discussants, while fellow discus-sants 'were rated low In the low and noorientation conditions.

The reason for this result could bethat two factors Influenced each person'sorientation responses. Festinger sug-

gested that the pressures of social realityand group locomotion act to increase the

24 Entmson, "Orientation Behavior on SmallGroup Consensus."

TABLE 12CELL AND MARGINAL MEAN ORIENTATION RESPONSE

BY ORIENTATION AND THREAT LEVELS

LowOrientation

HighOrientation Marginal

Low Threat 254 450 3.57High Threat 2.71 3.86 3.29Marginal 2.68 4.18

CONCILIATION AND PERRAL RESPONSES IS

TABLE 1$ANALYSIS OF VARIANCE SUMMARY TAUS Of ORIENTATION RESPONSE

Source di Mean Square F -Ratio

OrientatIvi (0) 31.50 19.756Threat (I) 1,14 0.72Sol (S1 3.50 2.190 X T 1.79 1.110 X S 2.57 1.61S X -r 0.07 0.04OXTXS 4.57 2.87Within 4 1.59

Significant beyond the .05 level.

probability of uniformity in a group.25In the high orientation condition, bothof these forces were at work. The subjectcould be expected to select high orient-ing statements because his social realitydemanded them (i.e., the other "sub-jects" were selecting high orienting state-ments), and because they were obviouslymore likely to bring about the group(and, incidentally, bis own) goal, con-sensus and a monetary reward. In thelow orientation condition, on the otherhand, social reality pressures demandedlow orienting statements (i.e., the other"subjects" were selecting such state-ments) and achievement of the goal,consensus, seemed highly improbable.Therefore, in the low orientation con-dition, social reality pressures workedagainst the selection of high orientationstatements, and group locomotion pres-sures were largely irrelevant.

Tables 14 through 16 ,show that sub-jects who received low orientation state-ments averaged 2.21 high threat state-ments in their responses, while thosewho received high orientation state-ments averaged only 1.04 high threat re-sponses. This is contrary to the reason-ing developed from the information andoutcome dependencies paradigm. Theobtained outcome could be due to thefrustration caused by the lack of infor-mation and direction in the low orienta-

25 Leon Festinger, 'Informal Social Communi-cation," Pochologicot Review, 57 (1950), 271.282.

tion discussions. This frustration, inturn, may have caused the subjects whoreceived low orientation statements toemit threatening statements. Though itmay be more logical to respond with in-formation which might maximize one'sreward, when there is consistent indif-ference on the part of the other mem'bets, a discussant seems more likely toemit threatening statements as a resultof his frustration.

Predictions 5 and 7 were contingentupon the support of prediction 8. Pre-diction 3 was not confirmed, and, as ex-pected, neither were predictions 5 and 7.

Predictions 4 and 6 were not con-firmed. This implies that there is littleor no differential effect of threat uponorientation in evoking either orienta-tion or threat responses. In the previousdiscussion, I suggested 'that when threatis very high a subject will respond withhigh threat and low orientation state-ments. Therefore, we would expectorientation responses to increase as ori-entation is increased under low to me-dium threat conditions; whereas, undervery high threat conditions, we wouldexpect little or no incr&se:in orienta-tion response as orientation statementsof the other subjects are increased. Con-versely, we would expect that for lowto medium threat conditions, as orienta-tion is increased threat response will de-crease; whereas for a very high threatcondition there is little or no decrease

16 SPEECH MONOGRAPHS

TABLE 14CELL MEANS OF THREAT RESPONSE

LowOrientation

HighOrientation Marginal

Male 1.79Low Threat 2.29 1.37High Threat 2.29 1,00

Female 1,46Low Threat 2.00 0.43High Threat

n =72.29 1,14

TABLE 15CELL AND MARGINAL MEAN THREAT RESPONSE

OF ORIENTATION AND THREAT LEVELS

LowOrientation

HighOrientation Marginal

Low Threat 2.14 1.00 1.57High Threat 2.29 1.07 1.68Marginal 2.21 1.04

TABLE 16ANALYSIS OF VARIANCE SUMMARY TAISLE OF THREAT RESPONSE

Source df Mean Square FRatioOrientation (0)Threat (r)Sex (S)0 X T0 X SS X T0 T X SWithin

1

1

1

48

19.450.161.450.020.452.160.881.43

13.61*0.111.010.010.311.510.61

Significant beyond the .03 level.

in threat response as orientation is in-creased .2$

Since sex was controlled in the experi-ment, I compared the threat and orien-tation measures of the verbal responsesof the male subjects with the female sub-jects. There was no significant differ-ence on these measures.

The verbal responses of the subjectswere also analyzed across time. Table 17indicates that the proportion of people

26To calculate the difference in response inorientation and threat across the five segmentsin discussion, 1 entploycd Cochran's Test, whichcan be viewed as'o a generalization of the Mc-Nemar twosample test, to calculate the signifi-cam trends. See, William L. Hays, Statistic,(New York: Holt, 1963), pr,. ,28-320, 531.360:Theodore J. Marr, "Conciliation and VerbalResponses as Functions of Orientation andThreat in Group Interaction," hiss. Universityof Iowa 1972, pp. 70.78, Appendix D.

who responded with high orientationstatements increased to a peak (78.5%)halfway through the discussion anddropped to a low ebb (37.5%) at theend of the discussion. It is a quadraticfunction of discussion segments, whereasthe proportion of people who respondedwith high threat statements is a quarticfunction of discussion segments. Sincethe bend at segment 4 is very slight,though significant, it is almost a cubicfunction. Table 17 shows that the pro-portion of subjects who responded withhigh threat statements rose from 89.3%at segment 1 to 57.1% at segment 2 anddropped to 14.3% at segment 3 and roseto 17.9% at segment 4 and continued torise to 33.9% at the final segment.

The distribution curve of the orienta-

CONCILIATION AND VERBAL RESPONSES

TABLE 17PERCENT Or SUOJEGIS RESPONDING WITH HIGH OPJENTATION

AND HIGH THREAT STATEMENTS

% of Subjects Discussion Session Segtnent

17

responding with: 1 2- 4 5

High orientation statements 64.3 73.2 78.5 71.4 37.5High threat statements 39.3 57.1 14.3 17.9 33.9

ziamseetleaA gn

don response indicates that there was arise in the use of high orientation afterthe discussion began which reached apeak toward the middle of the period.During the second half of the communi-cation period high orientation state-ments dropped off. To the experimentalsubjects it seems to have been more im-portant to get their orientation in dur-ing the first half of the period than dur-ing the second half. In fact, when it wastime to make the final decision, most ofthe subjects ceased to provide high orien-tation.

The distribution curve of the threatresponse indicates that the applicationof threat pulsates in frequency. This isreminiscent of the Scheidel and Crowellspiral model of communication feed-back. Threat is increased at first andthen threat is withdrawn to an almostnegligible point. Toward the end of thediscussion, an increased number of sub-jects Applied high threat again.

Though no theoretical prediction ofthis communication process was formu-lated, previous experimental studies byFisher, and Bales and Strodtbeck, havepostulated phasic models which suggestthese results.

CONCLUSION AND IMPLICATIONS FOR

FUTURE RESEARCH

This study confirmed the findings ofprevious researchers that high orienta-tion verbal behavior evokes a greaterdegree of conciliatory behavior than loworientation verbal behavior in a consen-susachieving context. In addition, Ifound that, though varying levels of

threat statements alone do not Influenceconciliatory behavior differently, the ef-fect of orientation is not independent ofthe level of threat statements. The in-teraction between orientation and threatis such that an increasing level of orien-tation increases conciliatory behaviormuch more when threat is high thanwhen threat is low. The insignificant ef-fect of threat statements on conciliationin this study could be due to subjects'failure to perceive the threat as extreme.Future studies may vary the level ofthreatening verbal behavior until offen-sive rather than conciliatory responsesoccur.

This study provides insight into thefruitfulness of two of the theories usedin explaining group communication. Atheoretical rationale based on Thibautand Kelley's reward and cost paradigmcan be used to make some predictionsabout group communication, but it is in-adequate in some respects and needs tobe complemented. The interaction oforientation and threat can be explainedby combining the drive hypothesis withthe reward and cost paradigm.

The results of this study indicate thatfemales tend to compromise more thanmales. Furthermore, males apparentlyare less logical and more emotional intheir conciliatory response to verbalstatements.

The verbal response pattern of an in-dividual is affected by the orientationand threat statements of the other mem-bers in a group. Again Thibaut andKelley's paradigm, though it forms a basisfor an explanation, needs to be com-plemented with other social psycholog-

38 SPERM MONOGRAPHS

ical theories. The concepts of social reality and group locomotion in Feitin .gees informal sodal communicationtheory are useful for understanding theorientation response. In a situationwhere there is insufficient orientation,people tend to respond with less orienta-tion than when there is sufficient orien.tation. Insufficient orientation also gen.crates more threat responses. This maybe due to the frustration experiencedwhen the orientation is insufficient toaccomplish the group goal.

This study also suggested that groupcommunication as a process across timecan be systematically investigated a3 acontinuous function of many variables.I found an oscillating function of threatresponses across time whose shape is verysimilar to the sine curve. The orienta.tion response curve is very similar tothe inverted U curve. Further studiescan be designed to investigate whetherthis cycle repeats itself when time is ex-tended or whether it retains this par-ticular shape,

T.0

THE EFFECTS OF SUBSTANTIVE AND AFFECTIVECONFLICT IN PROBLEM-SOLVING GROUPS

MAE ARNOLD BELL

THIS study uses a conflict model totest the processes of opinion modi-

fication which sometimes lead to con-, sensus. The apparent paradox of ex-

plaining such processes through conflictis resolved by examining the definitionof consensus given by A. Craig Baird, adefinition which stresses the process in-volved in reaching consensus, not theoutcome of unanimity:

Discussion at its best means continual weighingof the proposition, modification of it, even sub-stitution of a different proposal, until the ideasof the assembly coalesce.)

In order for ideas to coalesce, they mustbe juxtaposed. The implication that thearticulation and establishment of sepa-rate ideas is a necessary preliminary con-dition to reaching a consensual agree-ment has not been sounded as clearly asthe importance of achieving the goal ofunanimity. The focus of this study wasnot on consensus as an end product, buton the process across time defined byBaird which theoretically culminates inhigh quality solutions arrived at bygroup interaction.

A number of studies have attended tothe concerns of this study: (1) the kindsof interaction most likely to result in a"correct" or, sometimes, a consensus so-

Ms. Bell is completing her doctorate at the Uni-versity of Iowa, where, in 1971.73, she will beassistant professor in the Rhetoric Program andDepartment of Speech and Dramatic Art. Shekindly wrote this paper to fit the availablespace. For further detail, see her forthcomingdissertation.

I A. Craig Baird. Public Discussion and De-bate (Boston: Ginn and Co., 1937). p. 337.

SPEECH MONOGRAPHS, Vol. 41, March 1974

lution;2 (2) the contributions of sub-stantive and affective conflict to thesekinds of solutions;s and (3) across time,the ways in which verbal contributionsinteract to produce predictable sequencesof interaction.4

The rationale called for a study ofgroup problem-solving in which the in-dependent variables of substantive andaffective verbal input are controlled inorder to measure both the individual

2 For example, John K. Brilhart and LureneM. Jochem, "Effects of Different Patterns onOutcomes of Problem-Solving Discussion," PUPnal of Applied Psychology, 48 (1964), 173.179;Carl E. Larson, "Forms of Analysis and SmallGroup Problem-Solving," Speech Monographs,36 (1969), 454; and Irving L. Janis, Victims ofGroupthink (Boston: Houghton Mifflin Co.,1972).

Most noteworthy are Paul E. Torrance,"Function of Expressed Disagreement in SmallGroup Processes, Social Forces, 35 (1957), 318;Harold Guetzkow and John Gyr, "An Analysisof Communication in Dedsion-Making Groups,"Human Relations, 7 (1954), 367.382; DennisGouran, "Variables Related to Consensus inGroup Discussion of Questions of Policy," Diss.University of Iowa 1968; Thomas J. Knutson,"An Experimental Study of the Effects of Ori-entation Behavior on Small Group Consensus,"Speech Monographs, 39 (1972), 159-163; Theo-dore J. Marr, "Conciliation and Verbal Re-sponse as Functions of Orientation and Threatin Group Interaction," Diss. University of Iowa1972; and Dale G. Leathers, "Process Disrup-tion and Measurement in Small Group Com-munication," Quarterly Journal of Speech, 55(1969), 290.

4 Robirt F. Bales and Fred L. Strodtbick,'Phases in Group Problem-Solving," Journal ofAbnormal and Social Psychology, 47 (1951), 495;Laura Crowell and Thomas M. Scheidel, "Cate-gories for Analysis of Idea Development In Dis-cussion Groups," Journal of Social PsyChOlOgy,54 (1961), 155.168; Thomas M. Scheidel andLaura Crowell, "Feedback in Small Group Com-munication," Quarterly Journal of Speech, 52(1966), 278; Aubrey Fisher, "Decision Emer-gence: Phases in Group Decision Making,"Speech Monographs, 37 0970, 53-66; and Mare.

20 SPEECH SIONOCRAPHS

verbal contributions of subjects andthe process of reaching a solution. Sucha study faces the dilemma that plaguesmost group communication studiesrig-orous control of communication vari-ables creates atypical situations, the re-sults of which are questionable in gen.eralizing to realistic communication con-texts. On the other hand, the lack ofcontrol existent when groups interactwithout any external restraint results inconfounded and sometimes contradic-tory causal explanations.

The basic question was how to createcredible small groups and still main-tain relatively complete control overtheir interaction. As Mart had suggested,game simulation offered at least the ad.vantage of control. For example, by hav-ing fOur subjects believe that they areplaying a game with the objective ofchoosing a correct solution, but restrict-ing their communication to notes sentto each other, the experimenter can con.trol every message each subject receives.Of course, nonverbal communicationwould have to be eliminated; the sub-jects would not be permitted to see orhear each other.

Mare accomplished such control byhaving his subjects separated in cubicleswith art opening for passing notes pre-pared in advance by the experimenter.There are some obvious drawbacks tosuch a minimal social situation. One isgeneralizability. Another is the credibil-ity of the experimentdo the subjectsbelieve they are playing with the per-sons in the other cubicles? A final limi-tation is that forced response paradigmrestricting the subject's responses toone of four messagesmay not be an ap-propriate measure of his or her com-municative intentions. The messageswere constructed by someone else. Theproblem remained of maintaining rigorous control over what messages the sub-

jects received without closing them upin cubicles and without controlling theresponses they sent.

That problem was resolved by usingthe University of Iowa's Computer As-sisted instruction Laboratory, which isequipped with cathode ray tube termi-nals. Each terminal resembles a portabletelevision set which has been equippedwith a typewriter keyboard. The subjectcould see the messages from the otherplayers (which would be controlled bythe experimenter) on his or her screen.The subject could also type in a messageto the other group members whichwould be displayed on his or her screenas well as presumably on theirs. Thesubject could type in whatever messagehe or she wanted to communicate. Thelab has sixteen, terminals, making itfeasible to run sixteen subjects eachhour.

METHOD

The experiment consisted of a simu-lated discussion of a problem designedby Norman R. F. Maier and Allen R.Solem, often referred to as the horse-trading problem: "A man bought a horsefor $60 and sold it for $70. Then hebought it back for $80 and sold it againfor $90. How much (lid he make or losein the horse business?"5 When they had353 college students solve this problem,15.8 percent selected the correct answer"made $20." The four incorrect an-swers given by over half their subjectswere "lost $10," "broke even," "made$10," and "made $30." This study elimi-nated "lost $10" as the least defensibleand proceeded with the other three in-correct answers and the correct one.

Norman R. F. Maier and Allen R. Solem,"The Contribution of a Discussion Leader tothe Quality of Group Thinking: The EffectiveUse of Minority Opinions," in Group Dynamics:Research and Theory, ed. Dorwin Cartwrightand Alvin Zander, 1st ed. (White Plains. NewYork: Row, Peterson, 1953), pp. 561-572.

SUBSTANTIVE AND AFFECTIVE CONFLICT 21

In order to control all the verbal mes-sages a player received from the otherplayers, a complete set of stimulus mes-sages defending all four possible answerswas constructed for each of four conflicttreatment conditions: high-substantive/high-affective, high-substantive/low-af-fective, low-substantive/high-affective,and low-substantive/low-affective. Forexample, the following is a set of mes-sages fitting each combination of vari-ables in defense of "made $10."

High substantive/high-affective: Your logic iswrong. Stop moving the numbers around andlisten. Look, he sold the horse for a $10 profit,then he bought it back and lost that $10 be-cause it cost him $10 more than he sold It for.Now that's obvious he made a $10 profit. Ifyou can't see that, something is wrong withyou.

tbw-substantive/high-affective: Your logic is

t,rong. Stop moving the numbers around andlisten. It's obvious that he made a $10 profit. Ifyou can't see that, something is wrong with you.

High-substantive/low-affective: Look, he soldthe horse for a $10 profit, then he bought itback and lost $10 because it cost hint $10 morethan he sold it for. He then sold it again forSIO more than he bought it for. He made a $10profit.

Losubstantive/low-affective: He made a $10profit.

In the experiment, each subject wasled to believe he or she was communicat-ing via the terminal with three otherpeople in the room. Actually, each sub-ject received one of four sets of previ-ously prepared messages defending thethree solutions not selected, Subjectswere randomly assigned to treatmentconditions. Each subject made five re-sponses during the experiment,

The dependent variables to be ana-lyzed were the change to final correctsolution and characteristics of each sub-ject's five verbal responses. The latterwere analyzed not only for substantive,affective, and metadiscussional content,but also for changes across time. Becauseof space limitations results and interpre-

tation of correct solutions will not bereported here.

Two judges rated the verbal re-sponses.8 The operational definition ofa highly substantive response was as fol-lows: "substantive if it offers a mathe-matical procedure indicating how thesource of the response arrived at his orher answer." A response was defined ashighly affective, "if it contains words ofantagonism toward other people or to-ward their ideas." The metadiscussionalcontent was defined as "metadiscussionalif the statement makes a comment and/or a suggestion about the discussion pro-cedure itself, as opposed to the problembeing discussed." The combined reli-ability ratings for each of the three ver-bal variables were as follows: substantive.98; affective .81; metadiscussional .84.

Sets of hypotheses were generated forthe substantive, affective, and metadis-cussional content of the verbal responses,based on the theories of social compari-son and group locomotion. Social com-parison theory claims there is a tendencyin human beings to evaluate their be-liefs and behaviors: when physical re-ality checks are not available for theseevaluations, the person will use otherpersons as points of reference.T Thetheory is particularly applicable to situa-tions not experienced before, situations(as in this experiment) for which no in-dividual norms of behavior exist,

Hypotheses concerning substantivecontent:

la. If three of four members make highly sub-stantive statements, the remaining memberwill contribute similarly. Or conversely, ifthree of four members make lowsubstan-tive statements, the remaining member willmake tow - substantive statements.

a I am indebted to Katrina Simmons andConnie Swank for their time and performanceas judges.

1 Leon Festinger, "A Theory of Social Com-parison Processes,' Human Relations, 7 (1954),117.110.

22 SPEECH MONOGRAPHS

lb. The mean substantive content of responsesin the four experimental conditions will bein the following descending order: high-substantive/low-affective > high-substantive/high-affective t> low-substantive/high-affective > low-Iabstantive/low-affective.

Hypotheses concerning affective ver-bal behavior:

2a. If three of four members make highly al-teethe statements, the remaining memberwill contribute similarly. Or conversely, ifthree of four members make low-affectivestatements, the remaining member will alsomake low-affective statements.

2b. The mean affective content of responses inthe four experimental conditions will be inthe following descending order: high-affec-tive/low-substantive > high-affective/high-substantive > low - affective /low- substantive>low-affective/high-substantive.

The hypotheses for metadiscussionalcontent are based on group locomotiontheorypressure for uniformity is de-rived from movement of a group towardits goa1.8 Verbal statements character-ized as procedural suggestions may beexplained by the individual's attempt tofacilitate goal achievement, a motiva-tion stronger when the group is progress-ing toward the goal. Such group-ori-ented statements seemed more likely tooccur in discussions where negativelyaffective interpersonal conflicts are at aminimum. Hypotheses concerning meta-discussional verbal behavior:

3a. If three of four group members make state-ments expressing little, if any negative af-fect for any other member or the groupas a whole, the remaining member will bemore likely to contribute highly metadis-cussional statements. Or conversely, if threeof four members make statements express-ing negative affect for other group mem-bers, the remaining meaner will not con-tribute metadiscussional statements.

3b. The mean metadiscussional content of re-sponses in the four experimental condi-

s Leon Festinger, "Informal Social Communi-cation," Psychological Review, 57 (1950), 271 -282.

lions will be In the following descendingorder: low-affective/high-substantive > low-affective/low-substantive > high-affective/highsubstantive > high-affective/low-sub-stantive.

RESULTS

Both hypotheses concerning substan-tive content were confirmed. The high-substantive stimulus messages producedresponses with significantly higher sub-stantive content than did the low-sub-stantive stimulus conditions. Analysis ofvariance indicated that differences inthe mean scores for the two substantivelevels were statistically significant (F =16.67, p <.01). Table 1 presents the cellmeans of the substantive scores by sub-stantive and affective levels. The order

TABLE 1CELL, MEANS OF SUBSTANTIVE SCORES

ON SUBSTANTIVE AND AFFECTIVE LEVELS

Low-Affective High-Affective

Low-Substantive 3.34 3.91High-Substantive 7.05 5.10

of the mean substantive content wasalso confirmed: high-substantivepow-af-fective > high-substantive/high-affective> low-substantive/high-affectiVe > low-substantive/low-affective.

The analysis of the affective contentof verbal responses confirmed one of thetwo hypotheses. Analysis of variance in-dicated that the differences in the meanscores for the affective levels was statisti-cally significant (F = 20.49, p <.01).The high-affective stimulus statementsproduced responses with significantlyhigher mean affective content than didthe low-affective stimulus messages. Themeans appear in Table 2. However, the

TABLE 2Cat MEANS OF AFFECTIVE SCORES

ON SVIISTAN1IVE AND AFFECTIVE LEVELS

Low-Affective High-Affective

LowSubstantive 2.60 4,37High-Substantive 2.94 3.98

SUBSTANTIVE AND AFFECTIVE CONFLICT 23

order of affective content in the low-af-fective conditions was reversed from theorder predicted. The data revealed thefollowing descending order: high-affec-tive/low-substantive > high-affective/high-substantive > low- affective /high-substantive > low-affective/ low-substan-dye.

Neither of the predictions for metadis-cussional content was confirmed. Theanalysis of variance did indicate statisti-cally significant differences in the meanmetadiscussional scores of the affectivelevels (F = 8.97, p <.01). But contraryto the hypothesis the high-affective stim-ulus conditions produced significantlyhigher mean metadiscussional contentthan did the low-affective stimulus con-ditions. Metadiscussional content pro-duced by the four stimulus conditionswas in the following descending order:low - substantive /high - affective > high-substantive/high-affective > low- substan-tive /low- affective > high-substantive/low-alfective. Those results are in Table3.

TALE 3CELL 'MEANS OF METADISCUSSIGS4 SCORESON SUBSTANTIVE AND AFFECTI%E LEVELS

Low-Affective High-Affective

Low-Subs lantive 3.06 4.10High-Substantive 2.92 3.37

To calculate the difference in trendsof the substantive, affective, and metadis-cussional scores across the five time seg-ments, formulas were solved for the lin-ear, quadratic, cubic, and quartic com-ponents of each of the three dependentverbal variables and analysis of variancewas conducted for each component oneach variable. The trend analysis of thisstudy differs substantially from the phaseanalyses of Bales and Strodtbeck, Schel-de) and Crowell, and Fisher. This study

approaches the process of communica-tion as a continuous function of manyvariables. The analyses of variance forthe trend components of the substan-live content indicated that the differ-.ences in the means for the linear trendwere statistically significant (F = 10.49,p <.01). A graph of the mean substan-tive scores across the five time segmentsrevealed that the substantive content ofth.1 responses tended to decrease slightlyover time. The analyses of variance ofthe affective content also indicated sig-nificant differences in the means for thelinear trend (F = 27.30, p <.01). Agraph of the affective scores' means re-vealed that they increased during the dis-cussion. The analyses of variance of thetrend components for the metadiscus-sional scores also indicated a significantlinear trend (F = 8.46, p <.01). Themetadiscussional content of the responsesalso tended to increase acrossinteractions of ,dl three anal, 'end-fled substantive conflict as altcontinuing function affecting all thrtvariables.

CONCLUSION

This study tested how verbal charac-teristics, specifically substance and af-fect, of the majority of group membersmay affect the verbal characteristics ofanother group member. Moreover, thestudy's use of a computer simulation hasimplications for methodological re-search. Simulations on an interactivecomputer system offer a potential forgroup studies that needs further exploi-tation. The trend analysis of this study,as a departure from the emphasis oncorrelational data to explain processover time, suggests approaching the pro-cess of communication as a continuousfunction of many variables.

COMMUNICATION IN GAME SIMULATED CONFLICTS:TWO*EXPERIMENTS

THOMAS M. STEINFATT, DAVID R. SEIBOLD, and JERRY K. FRYE

,study

are many approaches to thestudy of conflict, several of which

are discussed in the forthcoming bookedited by Miller and Simons.' The ap-proach of our studies is called GameTheory, which originated with theclassic work by Von Newmann and Mor-genstern in the early 19405.2 For our pur-poses, the method which Game Theoryhas devised to describe the situation inwhich behavior is occurring is of greaterinterest than its predictions concerningrational man. This method is the gamematrix, which describes all of the pos-sible behaviors for all parties to a sit-uation and the outcomes or consequencesof each of the choices.

The description is useful to studentsof communication and conflict for fourmain reasons. First, it allows the gainsand losses of the parties to be specified,and second, it allows the subjects tomake behavioral choices which result inthe gains or losses. Most research on theeffects of communication is conductedin the social equivalent of a partial vac-uum. The messages are related only tothe nothere and the not-now. There areno behaviors to be engaged in which

Thomas M. Steinfait is assistant professor ofcommunication at Queens College of the CityUniversity of New York. David R. Seibold fsa doctoral student in the Department of Com-munication at Michigan State University. JerryK. Frye is assistant professor of speech com-munication at the State University of New Yorkat Buffalo.

I Gerald R. Miller and Herbert W. Simons,eds., Perspectives on Communication in SocialConflict (Englewood Cliffs: PrenticeHall, inpress).

2 John Von Neumann and Oskar Morgen-stern, Theory of Games and Economic Behavior(Princeton: Princeton U. Press, 1944).

SPEECH MONOGRAPHS, Vol. 41, March 1971

will produce real gains or losses to thesources or receivers of the messages intheir current situation. The game matrixprovides one means of adding the situa-tional variables of behavior and rewardto the context of an experiment. Thus,results of such experiments are morelikely, to generalize to situations inwhich communication functions as anessential determinant of behavioralchoices.

The game matrix also allows a thirdvariable to be studied: power. Any ma-trix can be constructed so that the re-wards a person receives in a situationare partly dependent on his own be-haviors, and partly dependent upon thechoices of the other person. It is ourcontention that much communicationbehavior takes place in situations wherethe source has the power to help or hitt-der the receiver in his progress towardcertain goals, and the receiver likewisehas the power to help or hinder thesource. One's behaviors in a powerlesssituation may bear little relationship toone's actions when either or both ofthe parties to a situation has limitedpower over the others. The fourth bene-fit derives from the problem of the re-lationship between the concepts of at-titude and behavior. It becomes unnec-essary to ask if attitudes actually predictfuture behaviors, since the behavioralchoices made in the game are a directindex of behavior when the rewards arereal to the subjects. If one is interestedin the attitude-behavior problem, thegame situation allows one to investigateit by asking subjects how they intendto play the game, and correlating this

COMMUNICATION IN GAME SIMULATED CONFLICTS 23

if both players make the competitivechoice, both will lose.

The name, Prisoner's Dilemma, is de-rived from one explanation of the gamegiven to persons who play it. SupposeP and 0 are two prisoners just appre-hended by the local police. They are sep-arated and told they have two choices:remain silent, represented by the C orcooperative choice, or confess and betraythe other, the D or defecting choice. Thetwo choices' are displayed in Figure 1.

The choice hinges on whether each cantrust the other to remain silent. For ifone confesses, he will go free and 'berewarded by the police while the otherwill receive a heavy sentence. If bothconfess, both will go to jail, but they willreceive relatively light sentences. Finally,if each remains silent, trusting the otherwill also do so, the police will have nocase, and both will be released after onlya few days in jail.

measure of attitude with the actual be-haviors evidenced in the game. But theresearcher can side-step the construct ofattitude completely if he wishes, andconcentrate his energies on the depen-dent variable of behavior.

The remainder of this article reportsthe results of several initial experimentson the effect of communication in gamesitnulated situations.3 These experimentsemployed full and open communicationin all conditions (no restrictions wereplaced on the possibilities for communi-cating either verbally or non-verbally)and did not manipulate communicationas an experimental variable. The resultsare compared with those of a priorstudy which did manipulate communica-tion.

Two types of games were used in theexperiments. The first was a Prisoner'sDilemtna game (PD) which simulates atype of interpersonal conflict. The sec-ond was a Creative Alternative game(CA) designed by the senior author tosimulate the type of situation in whichcollusive crime may occur.

Figure 1 displays a payoff matrix forthe first type of game, the Prisoner's Di-lemma, which was used in Experiment 1.Player 0 has two possible moves, C andI). If 0 plays C, he receives either +4or 2 units of reward, depending onP's choice. If 0 chooses D, he gets +6or0 units of reward, depending on P'smove. P's payoffs are similarly deter-mined by a combination of both players'moves. Given this matrix, C is the co-operative choice, and D is the competi-tive choice. Each player realizes that thecompetitive strategy is best for him per-sonally, .or it offers the possibility ofthe greatest gain with the least loss. But

3Sce 'Thomas M. Steinfatt and Gerald R.Miller. "Communication in Game TheoreticModels of Conflict," in Miller and Simons fora review of the literature on communicationin gaming studies.

0C

D

P

FIGURE 1: PD Matrix for Experiment I

Each of the studies reported here usedreal rewards. Several authors have dis-cussed the difference between real rewardand imaginary reward in game studies.4

4 Philip S. Gallo, Jr., "The Effects of Dif-ferent Motivational Orientations In a MixedMotive Game," Dist University of California

26 SPEECH MONOGRAPHS

A major point of these studies is thatwhen a reward has no real value to aperson it may be more interesting to in-vent a new game of maximizing thedifference between oneself and the otherplayer than to play in order tomize one's own reward. If generalizat6nfrom the gaming laboratory to non-laboratory situations is desired, the typeof situational difference imposed by adifference in reward conditions must betaken into account. Suppose I earn$10,000 a year and you earn $9,000. Oursupervisor gives you two choices. Eitheryou can reduce your salary to $8,000and mine to $7,000 or you can increaseboth of OW salaries by ten .percent oftheir current level. The choice seemsobvious. You might like to earn morethan 1, but not at the expense of a cutin salary. But in laboratory studies usingrewards of little or no real value to theperson, the results often do not reflectthis choice. Thus the cooperative-com-petitive measures used in studies of gamebehavior employing imaginary rewardsmay not directly generalize to non-lab-oratory settings unless those settings in-volve a strong motivation to maximizethe difference between persons.

In a study reported elsewhere, Stein-fatt used midterm examination pointsas real rewards in the PD game andfound a significant effect for communi-cation.3 He employed undergraduatestudents at the University of Michiganin three communication conditions over50 trials of the game and found thatconaihunication between the playersfrom trials I to 12 produced more co-operative responses than did a delayed

at Los Angeles 1963: Philip S. Gallo, Jr. andCharles C. McClintock, "Cooperative and Corn.petitive Behavior in MixedMotive, Games,"Journal of Conflict Resolution, 9 (1965), 68.78.

5 Thomas M. Steinfatt, "The Prisoner's Di-lemma and a Creative Alternative Came: theEffects of Communication hider Conditionsof Real Reward," Simulation and Games, 4(1973), 389-409.

communication condition in which corn-munication was allowed on trials 18 to25. Both immediate and delayed com-munication produced more cooperativebehavior than did a condition in whichortenunication was never allowed. Allubjetts in Steinfatt's experiments were

taller real reward conditions. The firstexperiment reported below attemptedto ref:di:ate the Michigan results and toinvestigate the thttertntial effects of realand imaginary rewards on cooperativebehavior in a PD game. It was expectedthat real rewards would produce morecooperative responses than imaginary rewards.

ExPERIMENT

Method

The game matrix used for ExperimentI is identical to that used by the Mich-igan subjects and appears in Figure 1.Subjects in Experiment I were 92 under-graduate students enrolled in speechcourses at the State University of NewYork at Buffalo. The purpose of the ex-periment was to determine the level ofcooperative responses in a PD game overGO trials under real and imaginary re-ward conditions. These conditions wereoperationalized as follows: For every 15points earned in the real reward condi-tions, the subject received one point onthe midterm examination. This connec-tion was made explicit in the instruc-tions to the subjects. In the imaginaryreward conditions, subjects were askedto imagine that they would receive onemidterm point for every fifteen gamepoints. Subjects played the game in aclassroom. The experimenter explainedhow to read a game matrix to the sub-jects and then seated each pair facingeach other. The game matrix was placedbetween the players and each subjecthad a pen and a score sheet in frontof him. Subjects were paired randomlyusing four classes totaling 92 students,

COMMUNICATION IN GAME SIMULATED CONFLICTS

28 pairs in the imaginary reward con-dition and 18 pairs in the real rewardcondition. All subjects completed 63trials of the game, but the last threetrials were not included in the analysisto avoid studying end-game effects. Sub-jects were informed before the start ofplay that it would be possible to splittheir pointswith the other player at theend of the trials if they so desired andthat they could talk with this partnerat any time. Except for these two condi-tions, the Buffalo experiment replicatedquite closely the real reward conditionsof the Michigan experiment.

Results and Discussion

No players in either condition askedto share ir rewards. This finding rep-licates thewhen thereof sharingthe possibilthe expertsharing oul

Michigan result obtainedwas the implicit possibilityn the situation (i.e., whenty was not made explicit byental instructions). Point

dde of the game itself doesnot seem to occur in PD games. The re-sults of ExAzriment I are summarized inTable I.

The difference in percentage of co-operative responses over all trials be-tween the two reward conditions is sig-nificant by Z-test for proportions (Z =1,812, p <.053). The number of pairswith 100% cooperation is interestingboth because this number influences thecooperative percentage and because it

may be compared for the two rewardconditions, Under imaginary reward coneditions from 10% to 36% of the pairswere responding completely cooperative-ly on any given trial block and three ofthe 28 pairs' made no competitive re-sponses for all 60 trials. This compareswith approximately 50% of the 18 pairsin the real reward condition who re-sponded completely cooperatively in anygiven trial block and the 6 pairs whomade no competitive responses through-out the trials. Thus, real rewards in afull communication PD game seem toresult in a level of cooperation signifi-cantly above that achieved under imag-inary rewards. A great portion of thisdifference is due to the number of pairswho form a 100% cooperative responseset. If only data from non-100% coop-erative pairs is analyzed, the differenceis still in favor of the real reward con-dition but is not significant. The effectof real rewards over imaginary rewardsseems to be to create more pairs whichrespond completely cooperatively, andto increase only slightly the level of co-operation in pairs that engage in at leastsome competitive behavior. It must beremembered that these statements applyonly to conditions of full communica-tion in a PD game when the possibilityfor side payments has been made explic-it. The Buffalo experiment (lid not in-vestigate situations of restricted com-munication.

TABLE 1COOPERATIVE RESPONSES FOR BLOCKS OF TEN Moos BY REWARD CONDITION,

Trials

Imaginary Reward Real Reward"NumbCr of Pairs Number of Pairs

Percent 100% Percent 100%Cooperative Cooperative Cooperative Cooperative

1.1011.2021-3031.4041.5051-60All trials

58626161686963

3578

10103

82817975848181

8to889

11

6

N = 28.*N = 18.

28 SPEECH MONOGRAPHS

A comparison of the results of Expert-molt I with Steinfatt's previous findingswith the Michigan subjects (see Figure2) indicates that the highest level of co-operation over all trials is found underconditions of real reward and full com-munication (81% in Experiment I and84% in Steinfatt's immediate communication condition). With imaginary re-ward and full communication (Expert.meet I) cooperation is 68%. With realreward but no communication cooperation drops to 32% (Steinfatt) which isnot different from the finding of 30%to 40% cooperation typical in PD gamesof imaginary reward and no communi-cation. Thus there seems to be no maineffect for real over imaginary reward.

Percentage ofCooperativeResponses

The existence of communication doesproduce an apparent main effect overno communication in a PD game. Inaddition, communication appears to in-teract with rewards to produce an evenhigher level of cooperation than isachieved with full communication alone.It is interesting that communication hasan effect in imaginary reward situationsbut that its major effect is reserved forthose situations where the rewards arereal.

To summarize, Experiment I replicated the results obtained by Steinfatt,and also found a significant difference inthe level of cooperative responses ob.tained under full communication, withthe real reward condition producing

43 b

FullCommunication Conditions

ImaginaryReward

ct.

NoCommunication Conditions

RealReward

aMichigan StudybBuffalo StudycTypical PD finding under imaginary reward

and no communication

FIGURE 2. Cross Experiment Comparison of the Effects of Communicationand Real Reward on Cooperative Choices in the Prisoner's Dilemma Game.

COMMUNICATION IN CAME SIMULATED CONFLICTS 29

higher cooperation than imaginary re-ward.

EXPERIMENT 11

Experiment 11 employed the creativealternative game (CA) used by Steinfattin his second experiment (Figure 3).°

A

P

FIGURE 3. Matrix forSteinfatt Creative Alternative Game

The CA game is an attempt to modelthe type of situation where collusivecrime may occur./ Collusive crime (crimewithout a victim) is a situation in whichtwo or more persons illegally enter into amutually beneficial agreement which in-creases the total payoff to the coalitionin one of two ways. (I) If the situationis defined as constant sum with "thehouse" as a player, the gain to the coali-tion produces a loss of like amount tothe house. (2) If the situation is viewedas variable sum the increased payoff oc-curs at no one's expense, resulting in-stead from increased utilities to bothparties. Examples of (I) might be the"spill money" offered by some stereoand high fidelity equipment manufactur-ers to salesmen and retail outlets topush products or to give favorable dis-

0Steinfatt, 403-406.7 See Steinfatt, 400.403, for a more complete

discussion of the CA game.

play space, and other examples of bribesand kickbacks as found in most of theWatergate related matters. Examples of(2), depending on one's point-of-view,might be the paying of a fee to a prosti-tute or buying a nickel bag of marl-Juana. The house is not usually repro-sented in a CA matrix but is the sourceof the actual payoffs in the (I) situation,If the mutually beneficial agreement isnot illegal, then the situation simulatedby the CA game is similar to the largeclass of human situations in which anadvantage will accrue to one of twopeople who are interacting if the otherchanges his or her current behaviorwhen he or she has no reason for doingso.

An important feature of a CA simu-lation is that the choice of the creativealternative must be self-generating. Itmust occur without any hint, suggestion,or encouragement by anyone other thanthe parties to the agreement. The valueof the payoff to both parties should besuch that at least one of the parties is ableto see the possibility of such an arrange-ment without receiving outside informa-tion that such a solution exists and with-out encouragement that either partymight wish to seek it. For such a situa-tion communication between the partieswill be of maximum importance.

.The CA game is quite different froma PD game. First, it is not symmetricsince the payoffs are not the same for Pand for 0. Secondly, either player in theCA game can guarantee himself a payoffof 4 units by making choice A for player0 or choice C for player P. At first glanceit appears that P is in a better positionthan 0 since P could get 20 units of re-ward and 0 can get only 4 units maxi-mum on any one trial. This advantagebecomes illusory on analysis since 0 hasno reason to ch000se B. If 0 chooses Ahe guarantees himself 4 units of rewardwhile if he chooses B he gets either noth-

30 SPEECH MONOGRAPHS

ing or loses 2 units. Thus 0 is goingto choose A. P may or may not see thisbefore the first move. If P concentrateson his own rewards and fails to analyzethe game from O's perspective before thefirst trial, then P may choose D for hisfirst response hoping that 0 will choose13, 0 or course chooses A, and after tiomore than three or four trials all P sub-jects in previous experiments (using nocommunication and Imaginary reward)extinguished on the D response.8

The CA game is a very boring, unin-teresting game when played under theseconditions. But suppose a new elementis added to the game. If subjects are al-lowed to communicate perhaps one ofthem will see a creative alternative tothe constant AC response pattern. Thethird' major difference of the CA gamefrom a PD game is that one of the cellscontains a joint total payoff which isgreater than the sum of the payoffsfor the obvious choice (AC) cell. Theexistence of this cell, the BD cell inFigure 3, has no bearing on the gamebehavior of the subjects when they can-not communicate. Would it have an ef-fect if they could communicate, especial-ly if they were under real reward con-ditions? Experiment H of the presentstudy attempted to answer this question.

But the implementation of full com-munication and real reward conditionswithin a game theoretic framework isstill not sufficient to explain real worldconflict; other variables such as per.sonality must also be considered. Herewe depart from' the bulk of game theoryresearch. Since game theorists are con-cerned with rationally conducted con-flicts and the purely structural featuresof games, game theory has emerged as a"depsychologized" decision theory, ac-cording to Rapoport? At the same time,

8 See Steinfatt, 401.0 Anatol Rapoport, "Conflict Resolution in

the Light of Game Theory and Beyond," in

because game researchers have been in-terested in how the game is played rath-er than who the player is, the relevanceof game theory to actors in real conflictsituations is often dubious. if a compre-hensive theory of cooperation-conflict isto be generated within a game-theoreticframeWork; personality variables mustbe accounted for.

The personality characteristic dogma-tism was chosen as an independentvariable in this study for two reasons.First, the results of previous game the-oretic studies suggest that traits such asabstractness-concreteness, tolerance ofambiguity, and dogmatism do affect co-operation-conflict' behavior, though theeffects have not been uniform;10 andsecond, in some of its features the nature

,of the game being played _is similar tothe Denny Doodlebug problem, usedby Rokeach to investigate the constructof dogmatism.n The problem cannot besolved until the assumptions and beliefsusually held about the ways animals andobjects move about are replaced withnew ones more relevant to Joe Doodle-bug's situation.

Subjects who are allowed to communi-cate while playing the CA game are ina situation similar to that of subjects at-tempting to solve the Doodlebug prob-lem. They must overcome specific beliefsabout what is possible in the game sit-uation and then develop new beliefs thatpermit a creative solution of the prob-lem. Given delayed communication inthe CA game, P develops a belief that

The Structure of Conflict, ed. Paul Swingle(New York: Academic Press, 1970), p. 1.

10 Daniel Druck man, "Dogmatism, Prenego-dation Experience, and Simulated Group Repre-sentation as Determinants of Dyadic Behaviorin a Bargaining Situation," Journal of Person-ality and Social Psychology, 6 (1967), 279.290;J. Gahagan, J. Horal, S. Berger, and J. Tedeschl,"Status and Authoritarianism in the Prisoner'sDilemma Game," paper presented at the meetingof the Southeastern Psychological Association,Atlanta, April 1967.

11 Milton Rokeach, The Open and ClosedMind (New York: Basic Books, 1960).

COMMUNICATION IN CAME SIMULATED CONFLICTS

O will always choose A, both from P'sobservation of the game matrix and fromhis observation of O's behavior, No mat-ter what the conditions of communica-tion, P also develops confidence in thebelief that the reward structure of thegame is irrevocably defined by the pay-off matrix presented to him. This beliefcorresponds to the real life assumptionthat, "Things are as they obviously ap-pear to be and cannot be changed." Athird belief inherent in most gamingsituations is that anything not specifical-ly allowed by the rules is specifically for-bidden. If P wishes to develop a crea-tive solution to the problem, he mustovercome all of these beliefs. The sameholds true for 0.

Since highly dogmatic persons takeconsiderably longer than their less dog-matic counterparts to solve the Doodle-bug problem and since the CA game hasbeen likened to the Doodlebug problem,it follows that highly dogmatic CA play-ers should be less successful than lessdogmatic players in arriving at the cre-ative solution. Previous research bySteinfatt seems supportive of this rea-soning.12 In a second experiment with48 undergraduates of the University ofMichigan playing 40 trials of the CAgame, he found that no pairs reached acreative solution when they were not al-lowed to communicate. With immed-iate and with delayed communicationunder real reward conditions he foundfrom 20 percent to 40 percent BD re-sponses. 'These BD responses came al-most exclusively from pairs involvingtwo less dogmatic players or a less dog-matic player in the P position. With nocommunication or with a highly dog-matic player in the P position he foundless than 5 percent (usually 0 percent) ofthe responses were BD responses. Thus,both communication and a less dogmatic

12 sEcinfatt, 406.

person in the P position were found tobe necessary, but not sufficient condi-tions for reaching a creative solution ina CA game. Once reached, that solutionwas fairly stable and did not disappearwhen communication was cut off. Thisresearch and the reasoning behind it ledus to the following hypothesis:

I. Highly dogmatic persons will be less likelyto achieve a creative solution (more thantwo consecutive BD responses) in a CAgame than will less dogmatic persons underconditions of full communication and realreward.

In addition to hypothesis 1, we wereinterested in any possible differences inbehavior between the Michigan under-graduates and inmates at a federal pris-on. We expected that both groups' be-havior in the CA game would be quitesimilar, despite differences in the demo-graphic characteristics 01 the subjectsand the nature of the real reward.

Method

The participants in this research wereall members of a college level speechcourse taught by the second author at afederal prison in Michigan during thefirst half of 1978. The prisoners, 21

males, were between the ages of 19 and80 years. Each student had at least a highschool education (or general equiValencydiploma) and several had completedsome college before their incarceration.Each student was acutely aware that hisfinal grade in this course and. otherclasses would help to determine (1) howsoon he might begin to travel to localcolleges for study-release time, and (2)in some cases, how much earlier parolemight be granted. In the thirteenthweek of class, forms containing a modifi-cation of Rokeach's 66-item DogmatismScale (Form D) were distributed to the21 students.13 The dogmatism scores ob-

13 Rokeach, pp. 913-415.

32 SPEECH MONOGRAPHS

wined on these sevenpoint items rangedfrom a low of 185 to a high of 292. Onesubject's form was discarded because healternately marked extreme ends of thescale. Based on these scores, subjectswere rank ordered from 1 to 20 (whererank I is the lowest dogmatism scoreand rank 20 the highest). Ten pairs ofsubjects were then formed In whichranks 1 and 2 were placed together, 3and 4 together, etc. Partners were there-fore paired with someone who was onlyone rank above or below their own.

The following week the students metfor a final exam. The instructor pro-tinted a copy of the test and explainedthat it consisted of 25 Identificationquestions worth one point each (25 per-cent of the final grade). The instructoradded that if the class agreed to partici-pate in a "learning exercise" they wouldnot have to complete each question onthe test. The purpose here was to inducea real reward condition. Based on hisperformance in the game, the studentwould be required to answer the 25 examquestions, less however many questionshe earned in the game. Hence, the stu-dent would be able to select those ques-tions for which he was best preparedand still achieve the maximum score byhaving those responses count more. Allclass members readily agreed to partici-pate.

The 20 subjects were divided into theten pairs drawn up after analyzing thedogmatism test data the previous week.To insure unformity, persons who hadthe lower score in each pair were as-signed the P position and each partnerthe 0 position.

A large matrix (like the one in Figure3) was drawn on the blackboard, and theinstructor took approximately ten min-utes to explain all possible moves andall possible rewards. No mention wasmade of the possibility or impossibilityof side payments. After several practice

trials the researcher specified the exactreward: for every twenty points wonafter the 37 game trials were corn-pleted,14 the student would have to doone question fewer on the final exam,Subjects were told that they were to hideeach decision until partners had markedtheir own decisions, never . to changemarks for any reason, and to do eachtrial independently and simultaneously.After each pair finished they were in-terviewed and asked: (1) Did they wish toshare any of their points with their part-ner? (2) Had their partner tried to in-fluence them to mark any particular de-cision at each trial, and, if so, with whatsuccess?Is

Results and Discussion

After 10 trials, the data from the paircomposed of the two subjects highest indogmatism were lost when P bitterlyabandoned the game after 0 would notlisten to his pleas for some degree of"cooperation." Player 0 was consistentlychoosing his A response and refused tochange his behavior to allow P to ob-tain the 20 points. Due to the intensityof this particular conflict, it was difficultto interview these two subjects, but itseems clear that P was asking 0 to en-gage in altruistic behavior and that theywere not discussing the possibility ofsplitting the 20 points from the BD cell.

14 The decision to have Ss perform 37 trialsin the Creative Alternative game stemmedfrom Steinfates (1973) finding that for playersallowed to communicate from the first trial tothe twelfth (but not subsequently) cooperativeresponses "dropped only slightly toward theend of the trials" (i.e., between trials 37 and SOthe percentage of cooperative trials droppedonly to 81 from 83).

15 The instructor explained the nature ofthe study to the students after the trials andpost experimental interviews were completed,and allowed each student to either 1) take as afinal grade the grade he had earned to thatPoint in the term; or 2) take the entire examwith his partner; or 3) work on the examalone, but omit questions based on the numberof points he had earned in the game.

CO/Ma/CATION IN GAME SIMULATED CONFLICTS 33

The subsequent analysis does not in-clude data from this pair except whereits inclusion is specified,

Only one pair achieved a creative so-lution to the game as defined by at leastthree consecutive BD responses. ThWpair was composed of less dogmatic play.ers who chose BD as their response forall 37 trials of the game, and in the postgame interview they requested a pointsplit resulting in 9 points for each play.er on each trial. No other players fez:'quested any point split. While severalless dogmatic players reported that theyhad considered, but rejected, the pos.sipility of BD responses and side pay.ments, none of the highly dogmatic pairsreported considering this possibility.While an occasional BD response oc-curred in the eight remaining pairs,these responses were isolated and seemto be due to chance maneuvering by theplayers rather than to a recognition ofthe possibilities of side payments. Theserandom BD responses occurred on theaverage of once every 11 trials for theeight remaining pairs. This is a higherrate than found with the Michigan un-dergraduates who made fewer than oneBD response per 40 trials in pairs whosecharacteristic response was not the BDresponse. Except for the difference inthe random BD response rate, the be-havior of the players in the Michigansample and the prison sample was quitesimilar.

Neither the Michigan study nor Ex-periment II was able to employ enoughsubjects in a real reward condition toconduct a meaningful test of significanceon the results with the CA game whenconsidered alone. By combining the re-sults of the two experiments, enough sub.jects are available to produce interpre-table results. Four pairs of highly dog.matic subjects and four pairs of less dog-matic subjects in Steinfatt's Michiganexperiment participated in communica-

tion conditions with real rewards thatare comparable to the conditions for Ex-periment I1.16 Three of the four less dog-matic Michigan pairs achieved a creativesolution while none of the highly dog-matic pairs did so. Combining these re-sults with the results of Experiment IIgives four of nine less dogmatic pairswith a creative solution and zero of ninehighly dogmatic pairs with a creative so-lution if the pair who stopped playingafter 10 trials is included. Using Fish-er's exact probability test (the hypergeo-metric distribution), the difference be-tween the highly= and the less dogmaticpairs is significant at the .0411 level. Ifthe data from the one highly dogmaticpair who almost came to blows and hadto stop playing is not included, the prob-ability is .0529. In either case, we wouldargue that these findings begin to pre-sent a convincing case that there is adifference in creative alternative gamebehavior between highly dogmatic andless dogmatic players. We regard thedata as generally supportive of our by.pothesis, but believe that more data areneeded before a stronger statement canbe made. Perhaps as convincing as thetest of significance is the fact that wehave yet to find a highly dogmatic pairwho seem to recognize the possibilitiesof side payments according to the postexperimental interviews. Several of theless dogmatic pairs who did not reacha creative solution in their game behav-ior reported the recognition of side pay-ments as a possibility, but for one reasonor another were not able to translatetheir thoughts into action.

CONCLUSIONS

The first experiment investigated be-havior in a PD game under full com-munication and compared the effects of

16 These are the HH and LL pairs in the ICand DC conditions of the second experimentreported by Steinfatt.

S4 SPEECH MONOGRAPHS

real and imaginary reward. Real rewardwas found to produce a higher level ofcooperative response than imaginary re-ward. When the results of Experiment Iare compared with the results of previousexperiments an interaction between re-ward conditions and communicationconditions is suggested. Under condi-tions 'of highly restricted communica-tion, real reward does not seem to pro-duce a level of cooperative responsewhich is substantially different from thatfound under imaginary reward condi-tions. But when full communication isallowed, real reward produces more co-operation than does imaginary reward.This apparent interaction is in additionto an apparent main effect for communi-cation across reward conditions. Thus,the significant effect for real over imag-inary reward found in Experiment I isbest regarded as a simple effect underfull communication rather than as a truemain effect. Ideally, these propositionsshould be tested in a single experimentrather than in the comparison of resultsacross different experiments.

The second experiment examined theeffect of communication on a situationwhich simulated the type of environ-ment which may result in collusivecrime: a desired goal is obtainable if oneperson can convince another that thegoal can, in fact, be attained. At leasttwo variables seem related to this processaccording to the results of ExperimentIL First, communication, the opportun-ity to exchange information concerningthe possibilities of the situation, is nec-essary. XVithout communication no cre-ative solutions occur. Second, the per-sonality variable dogmatism seems to berelated to the ability to achieve a cre-ative solution in a CA game. Does thismean that dogmatism is related to theprobability that an individual will en-gage in collusive crime? To the extentthat dogmatism is a measure of a per-

son's openness to new information andto new ways of thinking, perhaps it isrelated. The variable of dogmatismseemed relevant for inclusion in Experi-ment II due to the similarity of beliefchange processes involved in the solu-tion of both the CA game and the Den-ny Doodlebug problem. Persons low indogmatism may be more successful incompleting a belief change process whichprecedes collusive crime than are per-sons high in dogmatism.

But the action of the dogniatismvariable seems more complex than thissimple statement. Previous'resech hasindicated that dogmatism has itsistrong-est effect when very high - credible andvery low-credible sources are used.11Highly dogmatic individuals tend to actin accord with statements originatingfrom very positive (for them) sources andagainst statements from negative sourcessignificantly more often than their lessdogmatic counterparts.18 That is, highlydogmatic persons are more easily in-fluenced by persons they consider au-thority figures than are less dogmatic per-sons. It would seem that if the authorityfigures of a highly dogmatic person wereurging him to enter into collusion withthem that he would be more likely to doso than would a less dogmatic person inthe same situation. Yet less dogmatic sub-jects seem more capable of reaching acreative solution in a CA game than dohighly dogmatic subjects.

If authority figures were urging theperson not to engage in a particular col-lusion, we would expect the high-dog-matic to follow their advice more oftenthan would the low-dogmatic person. Wewould expect more collusive actions by

17 Erwin P. 13ettnghaus, Gerald R. Miller,and Thomas M. Steinfatt, "Source Evaluation,Syllogistic Content, and Judgments of LogicalValidity by High-and Low-Dogmatic Persons,"Journal of Personality and Social Psychology, 16(1970), 238-244.

18 Bettinghaus, Miller, and Steinfatt, 242-243.

COMMUNICATION IN CAME SIMULATED CONFLICTS 53

the less dogmatic subjects and fewer bythe highly dogmatic subjects. If the au-thority figures were urging collusion, itwould be difficult to predict a differencebetween subjects based on dogmatismsince the source argument would pre-dict more collusion with high dogmatics

while the Doodlebug argument of a be-lief change process would predict morecollusion by low dogmatics. Thus, fur-ther research on dogmatism in simula-tions of collusive crime is needed beforeany strong conclusions are drawn concerning its effects.

ARGUMENT IN NEGOTIATION:A THEORETICAL AND EMPIRICAL APPROACH

NANCY A. RETCHES and HARRIET B. HARRAI,

OUR concern is with the communi-cative process of negotiation as a

form of conflict resolution. Contempo-rary literature expresses the view thatthe role of communication in negotiationcannot be ignored. Communication re-searchers, however, have not identified,analyzed, and 4.4..teetated into a singlemodel the specific communication pat-terns that exist during bargaining. Weshall contend that viewing communica-tion in negotiation as an argumentativetransaction encourages heuristic analysisof that particular form of conflict inter-action. In this paper, we shall (1) pre-sent a paradigm to facilitate empiricalresearch, and (2) report the results oftwo empirical studies derived from thatparadigm.

Our concern for a communication-cen-tered theory of negotiation derives froma recognition that negotiation modelsoften ignore the interaction of the nego-tiators and instead focus on a determina-tion of outcome. Although these modelspurport to account for the behaviors ofeach negotiator, "most theories of bar-gaining do not give direct and explicitattention to the process of interactionbetween the parties."1 For example,some negotiation models view the bidsand counterbids of each bargainer as be-

Ms. Reiches and Ms. Harral began this Paperwhen both were graduate students at the Uni-versity of Colorado. Ms. Retches is now a doc-toral student at the Ohio State University, andMs. Harral, having completed her doctorateat Colorado, resides in Chicago. They wish toacknowledge the help of David H. Smith in thedesign of the experiments reported.

I Martin Patchen, "Models of Cooperationand Conflict: A Critical Review," Journal ofConflict Resolution, 14 (1970), 392.

SPEECH MONOGRAPHS, Vol. 41, March 1974

ing independent of his opponent. Somemodels "see the action of each party asaffected in part by its perception of theother side's likely actions."2 At least onedescription of the possible settlementincludes the "probability that a givendemand will be acceptable to the otherside."3 Though many theories do notanalyze the actions of a negotiator as be-ing a reaction to his opponent's previ-ous moves, the study of such reactionsis imperative if negotiation research isto be theoretically productive to com-munication scholarship.

The inappropriateness of outcome-centered models has led to our beliefthat we must identify and analyze thepatterns of communication manifest innegotiation. Our claim that negotiationinteraction can most usefully be recog-nized as argumentation is an extensionof J. Sawyer and H. Guetzkow's descrip-tion of:

the core of what is generally taken as the centralprocess of negotiationreciprocal argument andcounter-argument, proposal and counterpro-posal, in an attempt to agree upon actions andoutcomes mutually perceived as benefida1.4

Two initial observations about thenature of argumentation clarify ourreasoning. First, argumentation must bedifferentiated from a total conflict situa-tion. In argumentation, and negotiation,one "wins" relative to his own goals and

2 Patchen, 392.a Patchen, 392.4,Jack Sawyer and Harold Guetzkow, "Bar-

gaining and Negotiation in International Re-lations," in international Behavior: A Social-Psychological Analysis, ed. Herbert C. Kelman(New York: Holt, 1965), p. 479.

ARGUMENT IN NEGOTIATION

value system: satisfaction with the bar-gaining outcome does not necessarily im-ply crushing one's opponent. Negotia-tion, then, can be compared to a vari-able-sum game, a game that blends con-flict and cooperation, a game of mutualdependence. Thomas Schelling describesthese games in which:

though the element of conflict provides the dra.antic interest, mutual dependence is part ofthe logical structure and demands some kindof collaboration or mutual accommodationtacit, if not expliciteven if only in the avoid-ance of mutual disasters

Therefore, the distinguishing feature ofthe game is that both participants can"win" to some extent. For example, alabor contract must be acceptable toboth labor and management; both abuyer and, a seller must accept the termsof a sale or no transaction takes place.Each party wants the better of the deal,but the other party must also profit fromit if agreement is to occur.

A strict game theory approach is notsatisfactory if our goal is to generate acommunication theory of negotiation.The resolution of conflicting interestsmust be viewed as a process, not just theimplications of a set of outcomes. Forexample, what occurs during bargainingmay have ramifications reaching beyondthe explicit terms of the contract or legalsettlement. Again, outcome-orientedmodels are inadequate if a communica-tion theory of negotiation, and eventual-ly conflict resolution, is, to be gen-erated. Based on these assumptions, wecontend that our analysis of negotiationas an argumentative transaction beginswith the position at which much negoti-ation research ends.

Defining negotiation as a form of ar-gumentation is especially appealing be-cause of the close parallels between ar-

3 Thomas C. Schelling, The Strategy of Con-flict (Cambridge: Harvard Univ. Press, 1960), p.83.

gument and general concepts of conflittresolution. Resolution of a conflictandargument is conflictimplies, the con-frontation of choices and some uncer-tainty about a set of alternatives. Byargument we mean not only a, set ofstatements with supporting evidencecomprising a claim, but a concern withthe "techniques allowing us, to induceor to increase the mind's adherence tothe theses presented for its assent."e Inbargaining, such assent is crucial if a de-cision is to be reached. The notions ofa set of alternatives and of a choice-mak-ing process are also, critical. Once a setof alternatives has been articulated, thenegotiators employ forms of argumentin support of their most preferred posi-tion. One contender's arguments canincrease the number of choices availableto his opponent, can increase the uncer-tainty about some choices, and can fosterthe awareness that any of a range of al-ternative outcomes presents a more de-sirable solution than no agreement atall.?

In the same way that each bargainermakes choices about possible outcomes,he makes choices about the kind of ar-gument he will present. The forces in-fluencing these choices warrant carefulexamination; similarly, the communica-tive outcomes of these choices demandobservation and analysis. The claimsthat each bargainer can advance are de-termined by what we call dimensionsof argumentdimensions that aid crit-ical evaluation of any argumentativetransaction, but that are especially perti-nent to the examination of negotiation.

Ch. Perelman and L. OlbrechtsTyteca,The New Rhetoric: A Treatise on Argumenta-tion, trans. John Wilkinson and Purcell Weaver(Notre Dame, Ind.: Univ. of Notre Dame Press,1969), p. 4.

/ What we refer to here is that a widerrange of alternatives increases the latitude ofacceptable choices available to each bargainer.By Increasing the scope of alternatives, st fol.lows that the chances are increased for reach.ing a settlement.

Sit SPEECH MONOGRAPHS

The importance of a dimensional analy-sis lies with the assumption that the in-tegrated product of the dimensions, rath-er than discrete sets of isolated argumen-tative factors or components, governsthe form of the interaction. That is, nosingle dimension has total Influence overthe claims advanced, over the eventualsettlement, Or over the communicativeramifications of the negotiation expert-ence; rather, the dimensions interact toproduce a unique communicative trans-action. The five dimensions we shall de-scribe are power, risk, compromise, pre-diction, and situation.

Power and risk can most easily be dis-cussed together, since an important re-lationship exists between them. Percep-tions of power, determined by the rela-tive rewards and costs in a situation, al-so help to decide the amount of risk thata negotiator may take at a given time.Power is measured by the negotiator'sresources and the extent to which his op-ponent is dependent upon those re-sources. In this mixed-motive interac-tion, though, power relations are notnecessarily one-sided. As Wally Jacob-son su;4:ests, "because 0 has power overP does not mean that P is devoid ofinfluence capacity over 0. In any rela-tionship, then, each person may havesome power over the other, one may haveall the power, or neither may have su-perior power over the other."8 Whenexercising power, a negotiator must con-sider what he is concomitantly risking.Consequently, the negotiator's inherentpower is limited as the potential costsof his noncompliance increase. Overtime, the negotiator becomes more de-pendent upon his opponent's resources;he takes more risk each time he refusesto accept an offer. This reward-cost/pow-er-risk relationship is also implied by

s Wally D. Jacobson, Power and Interper.sonal Relations (Belmont, Calif.: Wadsworth,1972), p. 21.

J. C. Harsanyi: "A realistic quantitativedescription of A's power over B mustinclude, as an essential dimension ofthis power relation, the costs to A of at-tempting to influence B's behavior."0 Inother words, the attempt to Influence Bhas attendant to it both the measure ofA's power and the measure of riske-tenttal -costsChat A willing tO incur.'

This sto:ests that functional powercan be altered by one negotiator's abil-ity to increase the consequences of non-compliance for the other negotiator;that is, the ability of A to force B intotaking a high risk. B determines his risk(and possible reward) by comparing hisreward-cost outcome for compliance withthat for noncompliancethe greater thecost of noncompliance, the greater isA's power over B. This analysis impliesthe observation that as power relation-ships are altered, so are the correspond-ing measures of risk. The integration ofrisk, and power as determinants of argu-ment is inevitable in negotiation. Thepower that each negotiator attributes tohis opponent, as well as the amount ofrisk an individual is willing or able totake, can significantly influence thestatements spoken across the bargainingtable. In mixed-motive interaction, theprocess of argumentation and the pro-duct of negotiation rely heavily uponshifts in power perceptions, not solelyand perhaps not at allupon the in-herent power of 'the contenders.

The process of compromise is distin-.guished by a willingness to consider al-ternative proposals. The contenders ini-tiate the session by presenting their onemost preferred position, called a max-imum disposition.lo They gradually

9 John C. Harsanyi, "Measurement of SocialPower, Opportunity Costs, and the Theory ofTwo-Person Bargaining Games," BehavioralScience, 7 (1962), 70.

10 This analysis istased on the model offeredby David H. Smith.

ARGUMENT IN NEGOTIATION 39

move toward other. alternatives untilreaching the point beyond which neitherwill move. This "final offer" is termed theminimum disposition point. The rela-tive positions of the minimum disposi-tions determine what type of compromisemust occur to reach agreement. In CaseI, the minimum dispositions of the twoparticipants overlap; negotiators, then,must compromise on the most appro-priate alternative from among a set of,many suitable ones. Negotiation in Case1, therefore, involves an identificationof the overlap and a decision on the mostmutually acceptable position within thatlatitude.

A situation that initially presentschoices equally unacceptable to bothparties or a situation in which the onlyproposals are mutually unacceptableleads to another form of compromise. InCase II the participants are deadlockedand are forced to seek additional strate-gies or alternatives. While the game the-ory approach does not explain this sit-uation to our total satisfaction, it doeshelp clarify it. Stevens explains that themutually unacceptable alternatives re-sult in an "avoidance-avoidance conflictsituation."

The avoidance-avoidance conflict choice situa-tion is inherently and generically of such anature that the game take-it-orleave-it mustcreate strong motivations to discover alternativeresponses. In this situation the individual can-not immediately make up his mind which goalto elect. He is in a behavioral equilibrium suchthat strategies other than those available in thetake-it-or-leave-it are psychologically necessaryif the game is to be an appropriate one forresolving the transaction.11

When such alternatives are revealed, theparties are involved in a collaborativeeffort to maintain conditions underwhich disagreement rather than termina-tion is possible. Gradually, however, the

it Carl M. Stevens, Strategy and CollectiveBargaining Negotiation (New York: McGraw-Ili% 1963), p. IS.

contenders must lessen disagreement toreach a solution. In so doing, they par-ticipate in debate, a form of argumenta-tion. In Douglas Ehninger's terms, theyare engaged in a critical and cooperativeinvestigation. Particularly fitting thisview of negotiation, Ehninger claims thatargument encompasses those situations_In which mutually exclusive, or nonco-tenable, positions present themselves.Both arguers present their perspectiveson the issue, and both may examine,probe, and correct the other's viewpoints.Hence, they produce a dialectic, movingtoward mutually acceptable concha-sions.I2

The notion of a dialectic implies afourth dimension of argument in nego-tiationprediction. The argumentativeprocess culminating in agreement is sig-nificantly dependent upon behavioralpredictions based on the expectations ofthe participants. Schelling analyzes thisposition and concludes that "the out-come of a bargaining process is to be de-scribed most immediately, most straight-forwardly, and most empirically in termsof some phenomenon of stabilized con-vergent expectations."13 The coordinatedchoice, or outcome, occurs via the argu-mentative dialectic discussed above.Through the dialectic, each participantseeks to increase the accuracy of hispredictions concerning his opponent'sposition. The process is in a sense self-supporting, for as each party seeks tolearn more about his opponent, he alsoseeks to conceal his own minimum dis-position, frequently attempting to indi-cate that it is nearer his maximum dis-position than is actually the case. Thestrategies or tactics used to maximizeeach party's own preferred position arebased on the predictions each party

12 Douglas Ehninger, "Argument as Method:Its Nature, Its Limitations and Its Uses," SpeechMonographs, 37 (1970), 101-110.

is Schelling, p. 114.

40 SPEECH MONOGRAPHS

forms. An arguer must predict how com-mitted his opponent is to his expressedposition when attempting to decidewhether he must present another alter-native, or if he can expect his opponentto make the first move.

The process is analogous to a gameof "chicken." The first expressed willing-ness to move from a position will betaken advantage of, and the person ex-pressing the willingness may "lose"; butit neither participant swerves from hisinitial course, both will surely lose. Psy-chologically, perhaps, the last person tomove is invested with strength. As Schel-ling has suggested, though, "commit-ments are not altogether clear; each par.ty cannot exactly estimate the costs andvalues to the other side."14 Therefore,strategies to conceal a minimum dispo-sition and strategies to reveal more de-sirable positions are essential. Estimatesof an opponent's commitment to vari-ous claims in negotiation can determinethe course of the argumentative dialec-tic, the risks a contender will be willingto take, the functional uses of power,and even the eventual settlement forwhich each participant can estimate thathe could do no better.

A final class of variables affecting ne-gotiators' behavior can be distinguishedas situational variables. As in all inter-personal interaction, the actual verbalexchange cannot adequately be analyzedwithout consideration of the context inwhich it occurs. As Sereno and Morten-sen observe: "Every social situationforms a pattern, a context, that governsthe ongoing flow and effects of interper-sonal behavior."15 Situational variables,then, refer to external conditions uncon-trollably (for the most part) imposed up-on interpersonal encounters. A multi-

14 Schelling, p. 39.15 Kenneth K. Sereno and C. David Morten-

sen, Foundations of Communication Theory(New York: Harper, 1970), p. 292.

tude of such contextual variables arecrucial determinants of what statements,offers, and behaviors are permissiblewithin bargaining. Throughout this dis-cussion we have alluded to several waysin which situation affects negotiationit may suggest what compromises wouldbe acceptable; it might aid the bargainerin making predictions about his oppo-nent's minimum disposition; it may as-sist contenders in attributing power totheir opponents; it might assume a rolein convincing a bargainer that a par-ticular risk is worthwhile.

One particularly relevant variable inthe external situation is the urgency ofthe issue. The degree to which a decisionmust be reached in a specified, prede-termined amount of time relates directlyto notions of dependency in power. Forexample, if a union is threatening astrike on a particular date, and management could not absorb the losses a strikewould entail, the time pressures wouldcertainly affect management's negotiationbehavior, would affect the risks manage-ment might take when costs are high,and may also affect subsequent satisfac-tion with the settlement. The effects ofa situation may be reflected in a shiftof minimum disposition regarding wagesor concessions concerning benefits, work-ing conditions, etc. In any case, laborgains a power advantage as a result ofthe bargaining context. This exampleserves to illustrate the outstanding rolesthat situation, and context assume as de-terminants of negotiation interaction.

The model of negotiation that weproffer is based on the belief that ne-gotiation is a mixedmotive interaction,and, as such, it utilizes the dialectic ofargument to form converging opinionsof a best solution. Argumentation, then,is the vehicle by which the conflict is re-solved.

In the following portion of this pa-per, we shall report the results of two

ARGUMENT IN NEGOTIATION 41

exploratory empirical studies. These ex-periments were designed with the ex-pectation of developing useful categoriesfor subsequent content analysis. A corol-lary intent was to determine empiricallywhether the dimensions that we have se-lected are indeed significant; that is, wedesired to answer the question: Does Aremanipulation of a given dimension havea significant effect on the negotiationprocess or outcome? While we recognizethat isolating particular dimensions foranalysis may mask significant interactionpatterns among all dimensions, we de-termined that it was first necessary todescribe fully each individual dimensionand then examine all dimensions in aninteractive experimental setting. Hence,we selected to explore two dimensionsin this investigative, category-generatingresearch. Experiment 1 examines predic-tionspecifically perceptions of min-imum dispositions; Experiment II dealswith urgency of the issue as a possibleindicator of subsequent negotiator satis-faction.

EXPERIMENT I

Experiment I was designed to exam-ine at least one aspect of the predictiondimension. We have suggested thatthrough the argumentative dialectic,participants seek to gain maximum in-formation about their opponents' posi-tions. Our model suggests, then, thatone's ability to predict may be relatedto bargaining outcome. This experimentinvestigated that issue with the hopethat a useful category for future cameltanalyses would emerge. We believed thatthe ability to perceive accurately an opponent's minimum disposition at a givenpoint during bargaining would be agood indicator of a negotiator's skills.The implication is that a potentially suc-cessful negotiator will make use of theknowledge he possesses in advancing fur-

ther arguments. Based on this reasoning,the following hypothesis emerged:

Negotiators who more accurately predict theiropponents' minimum dispositions will gainmore favorable settlements than negotiators whomake inaccurate estimates.

Since the experiment reported below involved mock out-of-court cases, "favor-ability" was operationalized in the fol-lowing fashion: for plaintiffs, settlementsbecame "more favorable" as they ap-proached the total amount for which thedefendant was being sued. Conversely,for defendants, settlements became"more favorable" as they approachedzero. The "more accurate predictor" ofa pair was the one with a lesser dis-crepancy between his prediction of hisopponent's minimum disposition and theopponent's own statement of his min-imum disposition.

Method

Subjects. Undergraduate students (N= 52) enrolled in speech communicationclasses at the University of Nevada, LasVegas, served as subjects.

Design and Procedures. Subjects wererandomly placed in dyads and then arbi-trarily assigned the role of plaintiff ordefendant. The experimenter briefly ex-plained and illustrated to the studentsthe concepts of maximum and minimumdisposition presented above and deter-mined that all subjects understood thoseconcepts.

Subjects then received a brief descrip-tion of a mock litigation in which thedefendant was being sued for $10,000following an automobile accident, Fif-teen minutes were allotted for bargain-ing. After five minutes, and again afterten minutes, the experimenter haltednegotiation and instructed subjects toindicate their own minimum disposi-tion and to make an estimate of theiropponent's minimum disposition. The

42 SPEECH MONOGRAPHS

minimum dispositions were to be re-corded in dollar amounts, not via astatement such as "He is willing to payfor my medical expenses." At the end offifteen minutes, subjects were instructedto record the amount of the cash settle-ment upon which they had agreed.

nesuits

It was predicted that the more success-ful negotiators would have made moreaccurate estimates of their opponents'minimum dispositions. The data ob-tained suggest that th..: hypothesis is ten-able.

Data were analyzed in the followingmanner: for each subject, a percentagediscrepancy ((Ad) measure was calculatedfor each estimate that had been made.Percentage discrepancy represented thedifference between a subject's estimateof his opponent's minimum dispositionand his opponent's own statement ofminimum disposition. The percentage ofthe difference was calculated on the to-tal amount for which the subjects werenegotiating ($10,000). For example, if aplaintiff estimated the defendant's min-imum disposition to be $3000, and thedefendant stated his own minimum dis-position as $1500, the resulting %dwould be 15%. The difference between$3000 and $1500 is $1500; $1500 is 15%of $10,000.

Percentage discrepancy measures werethen correlated with the dollar settle-ments. Plaintiff and defendant data wereseparated for the correlations, since set-tlement favorability was operationalizedin opposing directions for plaintiffs and

defendants. Table 1 reports the resultsof the correlations.

In all cases, correlation coefficientswere in the predicted directions. Only forTime 2 (two-thirds of the way throughbargaining), however, were they signifi-cant beyond .05. While there is a rela-tionship between estimation and out-come, this relationship becomes morepronounced as bargaining progresses.

Discussion

It is not possible from this investiga-tion to ascertain fully the relationshipbetween prediction and negotiating abil-ity, but several conclusions are suggested.

First, the dimension of prediction ap-pears closely related to the negotiatedoutcome. Although this experiment didnot directly study bargaining strategies,one may now posit further research ques-tions in this regard: What distinguishingcommunicative strategies are typicallyemployed, the success of which are di-rectly related to knowledge of the op-ponent's position? A corollary to thisquestion poses four alternatives, whichmay be specifically addressed in futurecontent analysis research.

(1) Do successful negotiators intentionally orconsdously employ communicative strategiesthat conceal their own minimum dispositions?or (2) Are successful negotiators merely adeptat gleaning their opponents' minimum disposi-tions from statements made during bargaining?or (3) Do unsuccessful negotiators snake com-municative "errors" that reveal their minimumdispositions, hence granting their opponents astrategic edge? or (4) Do unsuccessful negoti-ators fail to exploit opportunities to perceivetheir opponents' minimum dispositions?

TABLE 1CORRELATIONS BETWEEN %D'S AND SETTLEMENTS FOR PLAINTIFFS AND DEFENDANTS

Plaintiff PlaintiffVcd, Time I r/cd, Time 2

Defendant DefendantVA, Time l (fccl,Time2

Settlement .292 .489* .371 .51.6*

'Significant at .05.

ARGUMENT IN

These questions bring to the forefrontthe larger conceptual and theoretical is-sue of the relationship between predic-tive ability as an indicator of communi-cative ability and eventual outcome.While prediction cannot be the sole de-terminant of success, it is probably a de-terminant of the kinds of arguments thata successful negotiator marshalls.

Two, specific suggestions for further;_probative work in this area can be madehere. In an attempt to determine a pos-sible answer to the four alternative ques-tions posed above, one might pair nego-tiators with previous records of success.Content analyses of these sessions may inturn suggest which communicative andpredictive strategies consistently pointtoward success.

Second, such content analyses and sub-sequent subject investigation may alsosuggest whether successful negotiatorsare more conscious of the advantagesgained by making predictions and esti-mates. Successful negotiators may exhibitan awareness of an interactive, ratherthan a linear, view of the communica-tion process. Criteria for content analy-ses could be developed whereby coderscould determine the extent to which astatement is in response or reaction toan opponent's statement or offer.

A second major area that warrants at-tention relates to the experimental find-ing that the correlation between cAd andsettlement increases in the predicted di-rection as bargaining time increases. 9neexplanation for this occurrence may bethat early negotiation time is spent morein information exchange and less in of-fer and counter-offer. Certainly, as morecommunication occurs, more is revealedhence, better predictions can be made.An alternative explanation is that thesoon-to-be-successful negotiator had al-ready gained a strategic advantage bythe second time estimates were recorded.This advantage may well account for his

NEGOTIATION 43

superior knowledge of his opponent'sposition.

It is apparent, then, that predictionis a significant dimension of the coin-municative/argumentative process evi-dent in negotiation. Further researchshould reveal the entire range of thisdimension's theoretic and practical im-pact.

EXPERIMENT II

Experiment II was designed to exam-ine aspects of the argumentative dimen-sions we have termed situation. We sug-gested earlier that the urgency withwhich a settlement must be reached mayaffect the claims exchanged during bar-gaining. The demand to make rapid andnumerous concessions (moves away fromone's maximum disposition) probablyaffects the perceptions a negotiator hasfollowing the bargaining session. Thisexperiment examined subjects' variousperceptions of the settlement and ofthe negotiation process under varyingconditions of "urgency." The followinghypotheses were examined.

1. The more time subjects are given to reacha settlement, the more satisfied they will bewith the settlement itself.

2. The more time subjects are given to reacha settlement, the more satisfied they will bewith the process utilized to reach that settle-ment.

The experimenters surmised that whengiven a generous amount of time, i.e.,under less situational pressure, subjectswill believe that they were able to con-sider carefully all aspects of the con-flict. Under restrictive, short time limits,on the other hand, subjects will likelybelieve that they were forced into ahasty, unconsidered decision.

Two additional hypotheses consid-ered the relationship between "satisfac-tion with settlement" and the settlementitself, and the relationship between "sat-

44 SPEECH MONOGRAPHS

isfaction with process" and the settle-ment.

3. Regardless of the amount of time allowedfor negotiation, there will be a direct re-lationship between "satisfaction with settle-melt" and the amount of the settlement.

4. Regardless of the amount of time allowedfor negotiation, there will be a direct rela-tionship between "satisfaction with process"and the amount of the settlement.

For plaintiffs, we anticipated a positivecorrelation between satisfaction and set-tlement, for defendants, a negative cor-relation. It seems reasonable to expectsatisfaction to relate with outcomewhether or not the issue involved is"urgent"people who "win" are gen-erally happier than those who "lose."Anyone, however, should be morepleased when given sufficient time toanalyze and resolve a conflict.

Method

Subjects. Undergraduate students (N168) at the University of Nevada, Las

Vegas, served as subjects. Subjects wereenrolled in classes in the Department ofSpeech and the College of Hotel Admin-istration.

Design and Procedures. The experi-mental paradigm is similar to the one re-ported above: subjects were asked to set-tie a mock litigation out of court. In thisexperiment, however, subjects were notintersupted during bargaining to makeestimates of minimum dispositions.Rather, they bargained for the entiretime allocated to their experimentalcondition.

Subjects were randomly assigned todyads, and dyads were randomly assignedto a treatment group. All groups nego-tiated the same case from the same setof mock legal facts. The case involveda liability suit for damages suffered ina fall on the edge of the defendant'sproperty. The only difference betweenconditions was the amount of time al-

lotted for reaching a settlement. GroupI was given ten minutes, Group II twen-ty minutes, and Group III thirty min-utes. At the end of the specified time,subjects were instructed to record thenumber of dollars, between $0 and$10,000, on which they had agreed. In-structions prior to the opening of thenegotiation session explicitly stated thatthe case could not be taken to court andthat a settlement must be reached with-in the time limit.

Subjects were then administered anevaluate six-scale, bipolar adjectivetest. The instrument was designed tomeasure subjects' _satisfaction with theoverall negotiation experience. The satis-faction instrument was divided into twoindependent parts. Scales 1.3 rated sub-jects' perceptions of the settlement thathad been reached. The adjective pairsfair-unfair, good-bad, and successful-un-successful were employed. On the finalthree scales, subjects rated their percep-tions of the process in which they hadjust been engaged. The adjective pairspleasant-unpleasant, meaningful-mean-ingless, and interesting-boring were util-ized here.

The semantic differential, then, yield-ed two measurements that could be uti-lized in analyzing the data: a judgmentabout the cash settlement and a judg-ment about the process employed toreach this settlement. Finally, the dataallowed for correlative comparisons ofindividual satisfaction scores with thecorresponding settlements.

Results

Satisfaction scores were computed byassigning the values 1-7 (negative topositive) to the various intervals of thesemantic differential scales. Scales 1-3rated satisfaction with the settlement,scales 4-6, satisfaction with the process.Tables 2 and 8 include means and stand-

ARGUMENT IN NEGOTIATION

TABLE 2MEANS AND STANDARD DEVIATIONS FOR "SATISFACTION WITH SETTLEMENT" SCORES

45

10 min. 20 min. 30 min.

Mean 15.14 16.86 16.71S. D. 4.92 335 3.73

TABLEMEANS AND STANDARD DEVIATIONS FOR "SATISFACTION wmi PROCESS" SCORES

10 min. 20 min. 30 min.

Mean 15.38 15.50 16.61S. D. 410 5.70 3.61

and deviations for all three experimentalconditions.

Hypotheses I and 2 were tested by cal-culating one-way ANOVA's based on thesatisfaction means, Tables 4 and 5 showthe ANOVA results. Since a significantF-ratio was obtained for the "satisfac-tion with settlement" ANOVA, multiplecomparisons were calculated. The Tukeytechnique was chosen to conduct thepair-wise contrasts. As Table 6 indicates,two of the comparisons proved signifi-cant at .05.

Based on this analysis, it is apparentthat Hypothesis I was supported..Satis-

faction with the settlement increased sig-nificantly as the time for bargaining in-creased from 10 to 20 and from 10 to50 minutes, though an almost impercep-tible decrease occurred from 20 to 30minutes. In general it may be suggestedthat time allotted influences satisfactionwith the bargaining outcome. The lackof difference between Conditions II andIll may suggest that after an optimalamount of time, satisfaction scores reacha plateau.

Hypothesis 2 was not confirmed. Al-though there was a slight overall trendin the expected direction, the resulting

TABLE 4ANOVA: SATISFACTION WITH SETTLEMENT

Source ofvariation

betweenwithin

df MS F p

2165

56.03 3.31 p <.0316.93

TABLE 5ANOVA: SATISFACTION WITH PROCESS

Source ofvariation

between'within

di

2165

MS F p

25.75 1.37 p <.2518.79

TABLE 6MULTIPLE COMPARISONS FOR THE 'SATISFACTION WITH SETTLEMENT" DATA

Comparison q Simultaneous Confidence Intervals

= (-1.72) 1.72 ± 1.001 (-2.721, .719)X1X111 = (L75) 519* 115 1.- 1.001= (-2.751, .749)X1IX = (.03)

TIT.099 .03 ± 1.001 = (-1.031, .971)

',Significant at .05.

46 SPEECH MONOGRAPHS

F-ratio fell short of the critical value.Apparently, satisfaction with the nego-tiation process is not significantly af-fected by the amount of time permittedfor bargaining.

Hypothesis 3 was examined by cor-relating individual satisfaction scoreswith the corresponding settlements forall three conditions. Table 7 includesthese results.

The test for the final hypothesis con-sisted of correlating "satisfaction withprocess" scores with settlements. Thecorrelation coefficients are reported inTable 8.

All correlations obtained were in thepredicted directions. The number of sig-nificant coefficients suggests that bothHypothesis 3 and Hypothesis 4 were sup-ported. It is important to note, however,that opposing trends emerge when com-paring the two sets of results. The cor-relations for "satisfaction with settle.ment" are more pronounced under con-ditions with limited negotiation time.The "satisfaction with process" coeffi-cients increase with more -ample negotia-tion time.

In sum, the following predictions wereconfirmed by the statistical analysis: (1)

Overall subject satisfaction with the set-tlement is greater when time granted toreach the settlement is generous. (2)There appears to be a direct relationship

between "satisfaction with settlement"and the settlement itself. (3) Further,there appears to be a direct relationshipbetween "satisfaction with process" andsettlement quality. The prediction thatoverall subject satisfaction with processwould increase with time was not borneout by the experimental findings.

Discussion

The results of this experiment clearly,indicate that the differential effects oftime upon negotiators' perceptions areworthy of further analysis. Several con-clusions drawn from the experiment of-fer implications for a dimensional analy-sis of argument in negotiation that canprovide a framework and viewpoint forfuture research.

One significant conclusion may bedrawn by comparing the outcomes ofHypothesis 1 with Hypothesis 3. Whileoverall satisfaction with the settlementincreases with time, individual satisfac-tion with the settlement becomes less re-lated to the quality of the settlement.That is, the satisfaction scores increasewith time, but their correlation with thesettlement decreases with time. This sug-gests that while greater negotiation timeyields greater satisfaction, the reason forincreased satisfaction must be somethingother than settlement size. This appar-ent interaction is further supported by

TABLE 7CoRRE1ATtoss Buns r.F.N "SATISFACTION WITH SETTLEMENT" AND SETIrl VAI ENT

10 min. 20 min. 30 min.

Plaintiff .611* .520' .2

Defendant .636* .412° .363Significant at b5.

TABLE 8CORRH.ITIONS fivF FN "S Mir ACTION 01111 PROCESS" AND SE1-11.1.NIFNI

PlaintiffDefendant

10 min.

.321

.100*

20 min.

.565.4330

30 min.

.601.5120

'Significant at .05.

ARGUMENT IN NEGOTIATION 47

the results for Hypothesis 4: with greaterbargaining time available, "satisfactionwith process" relates more closely to thequality of the settlement. It seems plau-sible to conclude, then, that when a set-(lenient is reached hurriedly, settlementsize'is the overwhelming factor in deter-mining high satisfaction. When a settle-ment is reached under more leisurelycircumstances, however, satisfaction isdetermined primarily by analysis of theprocessor at least by some factor otherthan settlement size, In,other words, un-der Condition III, it was possible to bevery pleased with the process, yet un-happy about the agreement, Conversely,under Condition I, it was likely that sub-jects who "won" (received high qualitysettlements) were less satisfied with theprocess employed to yield agreement.

The most important implication ofthe findings offered here is the reaffirma-tion of our earlier claim that outcome-oriented models are inadequate. The re-sults obtained in this experiment clearlysuggest that content analyses ought to beconducted to comprehend fully the com-municative patterns responsible for theobserved relationships between satisfac-tion, time, and settlement size. A cursoryglance at the human dynamics of negotia-tion might only suggest that "winning"(if winning is defined relative to settle-ment size) determines the attitudinal re-sponse. As we noted earlier, though, oneonly "wins" relative to his own goals.This research suggests that one of thoseother goals or values concerns the na-ture of communication occurring duringbargaining; the ramifications of a negotia-tion session extend beyond the simpleterms of the agreement. Participants'subsequent attitudes and behaviors mayin part be determined by bargaining ta-ble interaction. For example, the resultsof a first encounter between two con-tenders may well influence their behav-iors in a second negotiation.

The failure to confirm Hypothesis 2,when compared with the correlation re-sults, also yields interesting implications.Although overall satisfaction with theprocess was relatively constant acrossconditions, that measurement correlatedsignificantly for all but plaintiffs in Con-dition I. This again suggests that morethan outcome must be examined, Twoquestions for future inquiry presentthemselves: (1) What interaction pat-terns occur that allow "losers" to behighly satisfied with the process by whichthey "lost"? (2) What interaction pat-terns occur that allow "winners" to bedissatisfied with the process by whichthey "won"?

A final consideration is the issue oftime effects and settlement size. Al-though this experiment did not explorethat relationship, a communication per-spective toward negotiation suggests thatsuch investigation may prove significant.If length of bargaining time does notaffect outcome, then advantages gainedthrough increased time are solely "hu-man" or extra-negotiation advantages;if the only goal is reaching a settlement,however, then participant. satisfactionmight readily be sacrificed in order toconserve time and reduce bargainingcosts. If increased time tends to createconsistent advantages for one party,then the opponent may be acting ra-tionally if he forces an early settlement.In any case, the relationship betweentime and settlement size should be ex-amined as an important aspect of thesituational dimension.

Although many other situational fac-tors may affect communication in nego-tiation, this experiment has uncoveredtheoretically meaningful claims regarding the urgency of the issue. Further,the categories developed here promise tobe useful in observing future negotiationsessions.

48 SPEECH MONOGRAPHS

CONCLUSION

The purpose of this paper has beentwofold: to present a communication.centered model of the negotiation pro-cess and to explore empirically somedimensions of that model. The resultsappear promising in light of the claimsconfirmed here and the notions for fur-ther investigation that have been expli.

cated. The task now is to exploit furtherthe observational categories that emergedherein and to continue the explorationfor additional categories. Perhaps in hay.Ing identified the communicative processof negotiation as argumentation, a foun-dation has been laid for a better theoretical and functional understanding ofthe negotiation process itself.

PERCEIVING COMMUNICATION CONFLICT

THOMAS J. SAME

HREE theoretical traditions areprimarily responsible for the cur-

rent social psychological interest in hu-man conflict. Group-training theoristshave contended that conflict is a natural,human occurrence, which emerges as in-dividuals attempt to cope with role in-securities, dependence, and the desire forsocial acceptance.' By manipulating re-ward structures, game theorists havebeen able to study the processes wherebyoutcomes are negotiated in competitivesituations.2 Consistency theorists haveinvestigated the effect of intra-communi-cator conflict, generated by counteratti-tudinal role enactment, on attitudestructures.3

These traditions are similar in atleast one respect. Each has promoted thestudy of conflict effects. In other words,researchers have contrasted the impactof high versus low conflict situations onepistemic processes, including attitudechange, and communicative processes,including negotiated outcomes to inter-

- action. Much remains to be learnedabout what causes conflict to occur andhow or when or if people differ in theirperceptions of conflict. For example, wedo not know what information process-

Mr. Same is assistant Professor of speech at theUniversity of Florida. He expresses thanks toCarl Wattenbarger for assistance in conductingthe experiment.

I See Leland P. Bradford, Jack R. Cibb, andKenneth D. Benne, eds., %'Group Theory andLaboratory Method: Innovation in Re-education(New York: Wiley, 1964).

2Sce Elton B. McNeil, ed., The Nature ofHuman Conflict (Englewood Cliffs: Prentice-Hall, 1965).

Robert P. Abelson, Elliot Aton-on, WilliamJ. McGuire, Theodore M. Newcomb, Milton J.Rosenberg, and Percy H. Tannenbaum, edsTheories of Cognitive Consiwency: A Sotarebook(Chicago: Rand McNally, 196E).

SPEECH MONOGRAPHS, Pal. Ii, March 1971

ing skills affect the perception of con-flict. Anti we do not know what infor-mational characteristics of the commun-ication environment are necessary to en-able an observer to judge that conflicthas occurred or will occur.

This paper reports findings of experi-ments conducted to test the adequacywith which information processingtheory predicts individual ability to per-ceive conflict and rate its severity. Thisrelationship is of particular interest tocommunication theorists because itplaces the study of conflict perceptionwithin an understandable context ofother related decoding behaviors.

Information Processing

The term information processing re-fers to the ability of an interactant tocomprehend and use the alternatemeanings associated with a stimulus andto integrate that information into hisconceptual system in such a way that itrelates to previously acquired data with-out dissonance or stress' Two systemicproperties are typically associated withthe processing of information: cognitivedimensionality5 and integrative rules.6Although many theorists have attempted

4 Harold M. Schroder, Michael J. Driver,and Siegfried Streufert, Human InformationProcessing (New York: Holt, 1967), pp. 3-7.

3Schroder et al describe dimensionality:'Dimensions are the units of conceptual func.tioning and represent the elements or 'content'of thought. Judgments, attitudes, decisions, orperceptions concerning a range of stimuli canbe based on fewor manydimensional unitsof information (p. 7)."

Rules are the conceptual tools which allowan Individual to combine and reorganize cogniLions. The greater the number of Integrativerules, the greater one's ability to utilize diverseinformation.

50 SPEECH MONOGRAPHS

to label, describe, and statistically prediet individual differences in informa-tion processing, two theories in particu-lar appear suited to the study of con-flict perception because of their treat-merit of the two aforementioned con-ceptual properties and their overall pre-dictive power.

Cognitive Complexity Theory

Cognitive complexity theory repre-sents a reconceptualization and exten-sion of George Kelly's "psychology ofpersonal constructs."' Kelly argued thatpersonal constructs (actually bipolaradjective pairs, such as good-bad, effec-tive-ineffective, like-dislike, etc.) are theconceptual tools whereby man is able toconstrue meaning and differentiateamong similar cognitive and perceptualobjects .g Each individual has a limiteduniverse of personal constructs which hemay draw upon as he attempts to inter-pret life events. Theoretically, the great-er the number of personal constructs anindividual is able to bring to bear inunderstanding events, the greater hisability to comprehend fully the implica-tions of an act and the higher his level ofinformation processing.

Bieri used the label "cognitive com-plexity" to describe the relative com-plexity of personal construct systems.°Complexity theory stipulates that highcomplexity interactants, when con-fronted with complex informationalstimuli (complex because of the novelty,inconsistency, difficulty, or ambiguity ofthe stimulus), are better able to compre-hend, retain, and perceptively apply theinformation toward some post- communi-cation judgment. Research has demon-

George A. Kelly, The Psychology al Per-sonal Constructs (New York: Norton, 1951).

S Contemporary literature refers to "personalconstructs" as "judgmental dimensions."

James Bieri, "Cognitive Complexity-Sim-plicity and Predictive Behavior," Journal ofAbnormal and Social Psychology, 51 (1935). 263.

strated that high complexity decoders,given complex informational conditions,are better able to predict the behavior ofothers,to form more elaborate and multi-valent personal impressions,11 engageless in leveling (the omission of detailin recall) and assimilative projection(the perception of others in terms ofself),I2 and generally glean more infor-mation from messagesla than do lowcomplexity decoders.

Further, research suggests that cogni-tive complexity has an important andmeasurable influence on conflict percep-tion. Tripodi and flieri found that highcomplexity subjects incorporated signifi-cantly more conflicting themes in astory-completion task and were signifi-cantly more certain of their judgmentsregarding conflicting information thanwere low complexity subjects.14 It wasreasoned that "cognitively complex Ssproject more conflicting themes becauseof their greater versatility in conceptu-alizing dimensionS of behavior."3* Thesefindings are consistent with results re-ported by Leventhal and Singer which

fo Bieri, 263; Howard Leventhal, "CognitiveProcesses and Interpersonal Predictions," Jour-nal of Abnormal and Social Psychology, 55(1957), 176-180.

11 Louis J. N'idorf and Walter II, Crockett,"Cognitive Complexity and the Integration ofConflicting Information in WO:ten Impres-sions," Journal of Social Psychology. 66 (1965),165.169; Paul S. Rosenkrantr and Walter H.Crockett, "Some Factors Influencing the As-similation of Disparate Informat,on in Impres-sion Formation," Journal of Personality andSocial Psychology, 2 (1965), 317.402.

12 Richard M. Lundy, "Assimilative Projectionand Accuracy of Prediction In InterpersonalPerceptions," Journal of Abnormal and SocialPsychology, 52 (1956), 33-38; Leonard Berko-wit', "Leveling Tendencies and the Complexity-Simplicity Dimension," Journal of Personality,25 (1957), 743-751.

12 Tony Tiipodi and lames Biol. "Informa-tion Transmission in Clinical Judgments as aFunction of Stimulus Dimensionality and Cog-niche Complexity." Journal of Personality, 32(1964), 119-137.

11 Tony Tripodi and James Bicri, 'CognitiveComplexity, PerceWed Conflict, and Certainty,"Journal of Personality,. 31 (1966), 144-153.

13 Tripod' and Rim, "Cognitive Complexity.Perceived Conflict, and Certainty," ISO,

PF:RCFIVING COMMUNICATION CONFLICT 51

indicate that high complexity decodersexpress a preference for information re-lated to the inner states of principals ina hypothetical incident, while low com-plexity decoders search for statementsregarding surface qualities.". Followinga similar line of reasoning, Press, Crock-ett, and Rosenkrantz predicted andfound evidence to support the conten-tion that high complexity individualsexperience significantly less difficulty inlearning conflicting (unbalanced) socialstructures than do low complexity indi-viduals.17 The tentative conclusiondrawn from these studies is that highcomplexity interactams, due to theirability to create and tolerate conflictingperspectives on human behavior, aremore sensitive to interpersonal conflictthan are low complexity interactants.

Conceptual Structure Theory

Conceptual structure theory is partic-ularly attractive to communicationtheorists because of the strong emphasisplaced on the role of the communicationenvironment in decoding behavior.Schroder, Driver, and Streufert theorizedthat one's level of conceptual structureand the complexity of the communica-tion environment interact to affect thelevel at which information can be pro-cessed.18 A number of different factorsact either to increase or decrease thecomplexity of the communication en-vironment, including information load(the rate and/or amount of informationwhich is made available to decoders),

DJ Howard Leventhal and Dasid L. Singer,"Cognitive Complexity, Impression Formationand Impression Change." Journal of P,rsonatily,32 (1964). 210-226.

51 Allan N. Press, Il'aitcs II. Crockett. andPaul S. Rosenkrantz, " Cognitive Complexityand the Learning of Balanced and UnbalancedSocial. Structures," Jourrtol of Mootiality. 37(1969), 541-553.

18 Lod of conceptual structure' refers tothe way an individual receives. stores, procesfes.and transmits information," Schroder et al.. p. R.

information diversity, novelty of the sit-uation, severity of possible outcomes,and subject involvement or interest isExcessive or inadequate conditions forany of these factors create a below opti-mum environment for the processing ofinformation.

The specific relationship between con-ceptual structure and environmentalcomplexity can be represented as twoinverted U-shaped curves, with the curvefor low complexity persons being sig-nificantly lower in the moderate condi-tion of environmental complexity thanthe curve for high complexity persons.As the complexity of the communicationenvironment increases toward excessivecomplexity, information processing leveldecreases to a point at which the infor-mation processing levels for both highand low complexity persons arc approxi-mately equal. In other words, concep-tual. structure theory predicts that theinformation processing level of highcomplexity subjects will exceed that oflow complexity subjects only undermoderate conditions of environmentalcomplexity.

Research has provided strong supportfor this thesis. Studies have shown thathigh complexity decoders, under moder-ate conditions of environmental com-plexity, ask for less and utilize more in-formation,20 receive more information,vare better able to form new concepts,22better perform tasks requiring decic.ion

to Schtoder et al., pp. 31-53.20 Siegfried Streufert, Peter Suedfeld, and

'Michael J. Driver, "Conceptual Structure, In-formation Search, and Information Utilization,"Journal of Personality and Social Psychology, 2(1965), 736.740

21 Peter Suetlfeld and Richard Hagen, "Mea-surement of Information Complexity: Con-ceptual Structure and Information Pattern asFactors in Information Processing," Journalof Personality and Social Psychology, 4 (1966),231-236.

22 Joan Rigney, lames Bled, and Tony Tri-poli, "Social Concept Attainment and CognitiveComplexity, "Psychological Reports, 15 (1964),)01-509.

52 SPEECH MONOGRAPHS

integration (the formulation of specificbehavioral strategies which are interre-lated23), and are better able to tract andintegrate conflicting informationli thanlow complexity decoders.

Evidence exists which suggests thatthese same variables also interactto influence individual ability to per-ceive and evaluate the severity of inter-personal conflict. Crano and Schroder in-vestigated differences between high andlow complexity persons in their abilityand method of resolving conflict.25 Theresults show that following a strongcounterattitudinal message low complex-ity, subjects felt a greater need to re-solve conflict, used more of the availableprocesses for reducing conflict, and usedthem in a more internally consistentmanner than did high complexity sub-jects. Crano and Schroder interpretedthe results to suggest that:

As the complexity of a person's Informationprocessing structure increases, the individualwill generate more degrees of freedom in deal-ing with diversity. Consequently, there willbe 'a greater number of resolution processesavailable to him, and these will not necessarilybe bound by the condition of internal con-sistency. That is, multidimensional and alter-nate integrative rules will operate at more com-plex levels.28

These findings are reinforced by re-sults reported by Streufert and Streufertwhich indicate that conceptual structureand failuresuccess interact to affect theattribution of causality for interpersonalconflict.22 After being exposed to either

as Siegfried Streufert, Susan C. Streufert, andCarl H. Castore, "Complexity, Increasing Fail-ure, and Decision Making," Journal of EOCrimenial Research in Personality, 3 (1969), 293300.

21 Schroder et al., p. 112.a5 William D. Crano and If. M. Schroder,

"Complexity of Attitude Structure and Processesof Conflict Reduction," Journal of Personalityand Social Psychology, 5 (1967), 110.111.

28 Crano and Schroder, 112.27Siegfried Streufert and Susan C. Streufert.

"Effects of Conceptual Structure, Failure, and

conditions of increasing failure or in-creasing success in coping with interper-sonal conflict, subjects assigned responsi-bility for the outcome. Results show thatlow complexity subjects adhered morestrictly to the predictions of attributiontheory (i.e., that attitudes toward groupmembers will become more favorablewith success, individuals will take morepersonal credit for success and less forfailure) than did high complexity sub-jects. These findings were interpreted asevidence that high complexity persons,able to perceive an incident from multi-ple perspectives, are less likely to credita single person or condition with the re-sponsibility for either success or failure,while low complexity subjects, respond-ing to situations in a unidimensionalfashion, tend to centralize credit orblame.

Hypotheses

Ample theoretical grounds supportthe claim that the complexity of an indi-vidual's cognitive system influences hisability to determine whether interper-sonal conflict has occurred or will occurand to assess its relative magnitude. Fur-ther, there is reason to believe that thecomplexity of the communication en-vironment acts either to dramatize orminimize differences in conflict percep-tion. One factor in the communicationenvironment is of particular interestinformation load. Since perceptions ofconflict are based often on knowledgeabout the persons involved in the con.Ilia, their relationship, and disparitiesin opinion, the amount of informationmade available to an observer should af-fect the ability to perceive and rate inter-personal conflict. Cognitive complexitytheory and conceptual structure theory

Success on Attribution of Causality and Inter.personal Attitudes," Journal of Personality andSocial Psychology, 11 (1969), 138-147.

PERCEIVING COMMUNICATION CONFLICT 53

are the sources for the three hypothesestested in this study.

1. Cognitive complexity and information loadinteract to affect conflict perception, suchthat high complexity subjects, under moder-ate conditions of information load, recordmore instances of interpersonal conflict thando low complexity subjects, given these samecommunicative conditions.

2. Cognitive complexity and information loadinteract to affect the ability to integrate dis-similar dimensions of information in form-ing conflict perceptions, such that high com-plexity subjects, under moderate conditionsof information load, combine more dissimilardimensions of information to produce per-ceptions of conflict than do low complexitysubjects, given these same communicativeconditions.

3. Cognitive complexity and information loadinteract to affect the ratings of conflict se-verity, such that high complexity subjects,under moderate conditions of informationload, assign significantly higher numericalratings in estimating the degree of conflictthan do low complexity subjects, given thesesame communicative conditions.

EXPERIMENT I

Method

Subjects were 92 undergraduates en-rolled in sections of an introductoryspeech communication course who vol-unteered to participate. They filled outthe Paragraph Completion Test devel-oped by Schroder, Driver, and Streufertto assess the complexity of individualconceptual structures.28 It was possibleto classify subjects into conditions of"high" and "low" complexity by com-paring each individual score with thegrand mean for all subjects.29 Each sub-ject then received data on members of ahypothetical family (including informa-tion on the age, birth place, physicalappearance, political affiliation, religiousaffiliation, educational background ofeach family member, etc.). Subjects were

ssSchroder et al., pp. 189-198.29 Means closely approximate means reported

by Streufert and Streufert, p. 141_ .

instructed (1) to survey the data on fam-ily members, (2) to pair pieces of infor-mation which represent a potentialsource of family conflict (e.g., mother isCatholic, father is Jewish), reportingeach data pair on the "Conflict RecordForm," and (3) to rate each combina-tion (on a seven-point scale) to indicatethe degree of conflict one might expectto result. Subjects were instructed to re-cord as many sources of conflict as pos-sible during the allotted fifteen minutes,and to use their own judgment in de-

,termining what factors interact withother factors to produce conflict.

Information load was manipulated byvarying family size, while providingtwelve items of information on eachfamily member. In the sub-optimum con-dition, subjects were presented a two-person family (husband and wife), in-cluding the twelve pieces of informa-tion on each. Optimum information loadinvolved a three-person family (father,mother, and daughter), while supra-op-timum embraced a family of four (fa.titer, mother, daughtei, and son).30

Three dependent measures were re-corded: (1) total number of data com-binations listed by each subject, (2)number of multidimensional combina-tions (cross-category pairs, such as thepairing of mother's place of birth withfather's political affiliation), and (3) rat-ings of the degree of conflict.

Results

A two-way multivariate analysis ofvariance was performed to test the hy-potheses. Results for total number ofrecorded conflicts showed that subjectshigh in complexity recorded significant-

so Calculations to determine what levels ofinformation load would constitute sub-, opti-mum, and supra-levels were based upon pre-liminary testing in the introductory speechcourse.

54 SPEECH MONOGRAPHS

ly more instances of interpersonal con-flict than did low complexity subjects(F = 7.78, df = 1/86, p <.01). As pre.dieted, significant differences betweencell means (see Table 1) for subjectshigh and low in complexity occurred inthe optimum condition of informationload (t = 2.06, df = 86, p <.05).$1 Not

TABLE 1Nivacs FOR TOTAL NUMBER OF RECORDED

CONFLICTS FOR EXPERIMENTS 1 AND 2

Sub. Optimum Supra-

HighExperiment 1

Complexity 10.43 1230 12.76Low

Complexity 8.34 9.00 11.46

Experiment 2High

Complexity 7.90 12.36 9.65Low

Complexity 5.20 8.64 6.44

predicted, however, was a main effectdue to information load (F = 8.87, df =2/86, p <.05), and the finding that thenumber of reported conflicts increasedwith increments in information load,such that low complexity subjects in thesupra-optimum condition reported sig.nificantly more instances of conflict thandid low complexity subjects in the sub-optimal condition (t = 1.99, df = 86, p<.05). Effects of the hypothesized con-ceptual structure by information loadinteraction were trivial.

Consistent with predictions, results in-dicated that conceptual structure andinformation load interact to affect thenumber of multidimensional (cross-cate-gory) perceptions of conflict (F = 3.28,df = 2/86, p <.05), such that within theoptimum condition scores for subjectshigh in complexity were superior toscores for low complexity subjects (t8.70, df = 86, p <.01). Unanticipated

See B. J. Winer, Statistical Principles in Ex-perimental Design (New York: McGrawHill,1962), p. 244 for computation of t based onthe error term derived from analysis of variance.

was the evidence which shows this samesuperiority 10 occur under sub-optimumconditions of information load (t =8.11, df = 86, p <.01) (Table 2).

TABLE 2MEANS FOR TOTAL NUMUlt Or

MULTIDIMENSIONAL CONFLICTS RtcORprzFOR Ex Palms:14s 1 AND 2

Sub. Optimum Supra.

Experiment 1High Complexity 2.57 2.90 1.76Low Complexity 0.92 0.60 1.75

Experiment 2High Complexity 2.40 2.82 2.00Low Complexity 0.30 0.36 1.22

Tests for differences in ratings of thedegree of perceived conflict revealed oneimportant result: a significant maim ef-fect due to information load (F = 6.21,df = 2/86, p <.01). The pattern of cellmeans for subjects low in complexity fol.lowed prediction with ratings of great-est magnitude occurring under optimumconditions. Within the low complexitycondition, significant differences were re-corded between sub-optimum and opti.mum conditions (t = 2.37, df = 86, p<.05). In contrast, among subjects highin complexity the mean for conflict rat-ings was lowest in the sub-optimum celland highest in the supra-optimum cell(t = 3.57, df = 86, p <.01). The hy.pothesized conceptual structure by in.formation load effect on conflict ratingswas only marginally significant (F2.74, df = 2/86, p <.10), though in thepredicted direction.

EXPERIMENT II

Purpose

In Experiment I, the pattern of cellmeans for total number of conflicts re-corded departed somewhat from pre-dictions derived from information pro-cessing theory. Specifically, it was pre-dicted that as the complexity of the

PERCEIVING COMMUNICATION CONFLICT 35

communication environment increasedtoward a supra-optimal condition, de-creases should occur in the ability toperceive conflict. Results, however, in-dicated that additional increases in thenumber of conflicts recorded occurredin spite of excessive information load.Two explanations are possible. First, en-vironment and information processinglevel may be related in a linear, notcurvilinear fashion as suggested by in-formation processing theory, such thatability to perceive conflict increases asinformation load increases. Second, it ispossible that information load in thesupraoptimum condition was not suffi-ciently excessive to test for the invertedU-curve relationship. Experiment II is areplication of the original experimentwith modification in sub- and supra-optimum treatments to provide even lessinformation in the former condition andmore information in the latter.

MethodSubjects were 59 undergraduates who

volunteered to participate. As in theprevious experiment, subjects were ad-ministered the Paragraph CompletionTest and levels of conceptual structurewere determined by comparing individ-ual scores with the grand mean for allscores. As before, information load wasmanipulated by providing families ofdifferent size. In the sub-optimum con-dition, however, the amount of informa-tion provided was reduced by six items,while the amount of information pro-vided in the supra-optimum conditionwas increased by twelve items. Informa-tion load in the optimum condition wasthe same as in Experiment I. Measureswere taken of (1) total number of con-flicts recorded, (2) number of multidi-mensional combinations of data, and (3)ratings of conflict severity.

ResultsTests for differences in the total num-

ber of conflicts recorded yielded a maineffect for both conceptual structure (F =12.80, df = 2/53, p <.01) and informa-tion load (F = 6.37, df = 2/53, p <.01).Comparisons of cell means showed thathigh complexity subjects outperformedlow complexity subjects in the optimumcondition, as predicted (t = 2.51, df =53, p <.01). Unexpected was the superi-ority of high complexity subjects ovtrlow complexity subjects in the supra-op-timum condition (t = 2.16, df = 53, p<.05). Scores in the optimum conditionfor both high (t = 2.97, df = 53, p <.01)and low complexity subjects (t = 2.29,df = 53, p <.05) exceeded scores in thesub-optimum condition, but failed todiffer significantly from scores undersupraoptimum information load. Al-though the order of cell means was inthe predicted direction, the hypothe-sized complexity by information load ef-fect was not significant.

Analysis of variance for the number ofmultidimensional data combinations re-vealed a conceptual structure main ef-fect (F = 9.59, df = 1/53, p <.01).Scores for high complexity subjects ex-ceeded scores for low complexity sub-jects in both sub- (t = 2.14, df = 53, p<.05) and optimum conditions (t =2.24, df = 53, p <.05). Other treatmenteffects were trivial.

TABLE 3MEANS FOR CONFLICT RATINGS

FOR EXPERIMENTS 1 AND 2

Sub- Optimum Supra.

Experiment 1High Complexity 3.71 4.00 4.89Low Complexity 3.54 4.50 4.17

Experiment 2High Complexity 3.50 4.00 3.63Low Complexity 3.40 4.36 4.22

The main effect of information loador. ratings of conflict severity evidencedin Experiment I was not replicated. Noeffects on these ratings, including thehypothesized complexity by information

SPEECH MONOOMPHS

load effect on conflict ratings, were sitnificant.

DISCUSSION

Two experiments indicate that con-ceptual structure and information loadare important factors in conflict percep-tion. Although the predicted interactioneffects were not consistently achieved,the pattern of cell means and resultsfrom subsequent planned comparisonssuggest that the postttlate of an in-verted II-shaped .relationship betweenenvironment and information process-ing level applies to conflict perceptionas well as to other data. Apparently, asinformation load increases toward anoptimum condition, subject ability toperceive conflict also increases. Whileadditional environmental Stress causessome decrement in ability,,these differ-ences are not substantial. The consistentfailure to achieve a statistically signifi-cant decrease in conflict perception asinformation load was increased to supra-optimum levels suxests a need for amodification in theory. It seems unrea-sonable to expect information process-ing level, under supra-optimum condi-tions, to regress to the same low level ofperformance which occurs under sub-optimum conditions, especially when theenvironmental variable is information.We can expect information deprivationto be a far more debilitating factor thanexcessive information. Although excess-ive information may cause some difficul-ty in processing information, it shouldnot create a highly counterproductivecommunication environment. It is alsopossible that the rate of decay in levelof information processing as environ-mental complexity increases toward su-pra-optimum conditions is much slowerthan the rate at which improvement isachieved through increases in load.

An important finding was the con-sistent superiority of high complexity

subjects, both in the perception of con-filet and in the ability to combine dis-similar dimensions of data in construingconflict, over low complexity subjectsunder sub- and optimum conditions ofinformation load. Apparently, inform-don deprivation has a more serious andrestrictive impact on the informationprocessing skills of low complexity sub-jects than of high complexity subjects.

The failure to account consistently forvariance in rating behavior can be ex-plained in two ways. First, the ability todetect conflict and the ability to assessthe magnitude of conflict may be twodifferent cognitive processes. Just be-

cause an individual is able to judge thatconflict has occurred or will occur dciesnot necessarily imply that he is able torate the severity of the conflict. Second,it is possible that the nature of the ex-perimental task may have 'nuked' theeffects of the independent variables. Ifindeed high complexity subjects aremore sensitive to conflict than tows, im-plying here an ability to judge ac-

curately the magnitude of conflict, highratings would only occur when all in-stances of conflict are severe. In these ex-periments, high complexity subjects mayhave rated severe conflict higher andmild conflict lower than did low com-plexity subjects, thereby achieving amoderate overall estimate of conflict.

It seems clear from these experimentsthat factors which constitute the com-munication environment and concePtgalstructure of decoders are important fac-tors to be considered in constructing atheory of conflict perception. Informa-don processing theory appears to pro-vide a suitable theoretical context intowhich conflict perception can be fitted.Additional research is required to de-termine the precise impact of communi-cation variables on the ability to identi-fy and evaluate interpersonal conflict.

A LITERARY ANALOG TO CONFLICT THEORIES:THE POTENTIAL FOR THEORY CONSTRUCTION

LAWRENCE J. CHASE and CHARLES W. KNEUPPER

rr HE study of human conflict has at-tracted an interdisciplinary corps

of scholars, resulting in a rich mixtureof approaches and perspectives. How-ever, the potential contribution of liter-ary analysis in this area has not beengenerally recognized by conflict theorists.We contend that literature, particularlythe novel, may serve as a source of datawhich can be a valuable tool for the ex-plication and analysis of human con-flict.

Connected with this belief are two im-portant implications: (1) existing con-flict theories should have literary ana-logs; and (2) literature can provide animpetus to the generation of "new" con-flict theory. In order to illustrate theproposed relationship, we have under-taken a case study of Part I of FyodorDostoevsky's Crime and Punishment.This author and work were selecteC1 fora variety of reasons: (1) the title of thenovel is stpxestive of conflict; (2) it isa recognized literary classic; (3) it pre-dates contemporary conflict models; and(4) Dostoevsky is a prominent literaryfigure of the "realistic school" who hasdrawn considerable critical attention.For example, I. A. Richards, in "TheGod of Dostoevsky," characterized theRussian novelist as a "prophet," "art-ist," and "teacher."' Stefan Zweig de-scribed Dostoevsky as

Mr. Chase is a doctoral candidate in the De-partment of Speech Communication at Bowl-ing Green State University. Mr. Kneupper hascompleted his doctorate there and resides InSan Antonio. Texas.

I I. A. Richards, "The God of Dostoevsky,"The Forum, 78 (1927), 88-97.

SPEECH MONOGRAPHS, Vol. 41, March 1974

the psychologist of psychologists. SinceShakespeare lived and wrote we have notlearned so much from any one as from Dostoeff-sky about the secret sources of the emotions andthe magic laws which govern their interactions;and just as Odysseus was the only mortal whoever returned from Hades and told us of hisexperiences there, so Dostoeffsky relates hisvoyages in the underworld of the souLa

Moreover, Dostoevsky has previouslyreceived some consideration for the sci-entific implications of his work. Fried-rich Nietzsche referred to Dostoevsky'swork as "the most valuable psychologi-cal material known to me."3 In 1916, anarticle entitled "Dostoevsky as a Psycho!ogist" by G. W. Thorn appeared in theLondon Quarterly Reuieto.4 In 1927,Sigmund Freud wrote an essay entitled"Dostoevsky and Parricide."6 And mostrecently, in 1971 Pavel Simonov, a notedSoviet psychophysiologist, wrote "Dos-toevsky as a Social Scientist."6 Such pre-vious attention indicates the importanceof the author and the potential for sig-nificant discovery.

To amplify the dimensions of this po-tential, we provide: (1) a brief summaryof the plot structure of Crime and Pun-ishment, with special emphasis on Part

2 Stefan Zweig, Three Masters (New York'Viking, 1930), p. 204.

Friedrich Nietzsche, "To Georg Brandes,"Selected Letters of Friedrich Nietzsche, trans.and ed. Christopher Middleton (Chicago: Univ.of Chicago Press, 1969), p. 327.

4 G. W. Thorn, "Dostoevsky as a Psycholo-gist," London Quarterly Review, 125 (1916), 177188.

5 "Dostoevsky. and Parricide," in The Cons.piece Psychological Works of Sigmund Freud,XXI, ed. James Strachey (London: HogarthPress, 1961), pp. 175.196.

6 Pavel Simonov, "Dostoevsky as a Social Scientist," Psychology Today, 5, No. 7 (1971), 59106.

58 SPEECH MONOGRAPHS

1; (2) an illustration of literary analogsfor contemporary conflict models; and(8) a synopsis of the major implicationsof this investigation.

Raskolnikov, the central character ofCrime and Punishment, suffers intensepsychological conflict over a decision toattempt to validate a partially developedanthropological theory. Drawing uponthe ideas of Hegel and Nietzsche, Dos-toevsky provides Raskolnikov with an"extraordinary man" theory. This theoryis predicated upon the division of hu-manity into two distinct (lasses, extra-ordinary and ordinary. The extraordi-nary man is distinguished by his abilityto utter "something new," while theordinary individual is merely a "louse."For the extraordinary man, social con-straints do not apply, and, when he per-ceives a desirable goal, any means usedto attain that goal are justifiable. Thus,the murder of an ordinary individual,such as an old pawnbroker, may be jus-tified as necessary to further the ideasof the extraordinary man. After ini-tially deciding to commit the murder,Raskolnikov's uncertainty and compas-sion cause him to hesitate, to vacillatebetween acceptance and rejection of the"extraordinary man" theory, but even-tually he commits the murder, and intense mental conflict continues. The in-tellectual and compassionate sides of hisnature exert strong influences upon him,and he finally confesses his crime.

Raskolnikov's dual personality servesas a manifestation of the two conflictingforces acting within him. From the timethe murder is conceived until his confes-sion to the police, the interaction ofthese forces is developed in elegant de-tail. F. I. Evnin's "Plot Structure andRaskolnikov's Oscillations" provides ananalysis of the major dilemmas with

which the protagonist had to cope: "Hisentire progression in the novel Is an un-interrupted, tortured 'change of phases'of his internal struggle."' The decisionto commit the murder will serve as thefocal point of this discussion.

Prior to the commission of the crime,Raskolnikov changes his mind severaltimes. The first such change occurs dur-ing the trial run, where he meets thepawnbroker at her apartment. Afterleaving the building, he enters the streetand cries:'Oh God, how repulsive) can I possibly, can Ipossibly . . no, that's nonsense, its tirlkulOusi'he broke oil decisively. 'How could such a bor-Able idea enter my tnindra

Once he decides to abandon his plan, hevisits a tavern where he meets the post-erty-stricken Semyon lVfarmeladov, Mar-meiadov recounts the story of his life,and tells of the misfortune of his daugh-ter Sonia, who was forced to engage inprostitution to prOvide money for thefamily. Raskolnikov accompanies Mar-meladov home and secretly leaves somemoney. The compassionate nature ofRaskolnikov is aroused by Marmeladov'stragedy, and his doubt in the validity ofthe extraordinary man theory is rein-forced:

'Well, and If I am wrong,' he burst out sudden-ly, 'if men are not really scoundrels, men ingeneral, the whole human race, l. mean, thenall the rest is just prejudice, imaginary fears,and there are no real barriers, and that Is asIt should bell

Raskolnikov then returns to his apart-ment and sleeps until late the next day.At this point, the author describes Ras-kotnikov's environment:

I F. I. Evnin, "Plot Structure and Raskolni-kov's Oscillations," Tvorchestvo Dostoevskogo,trans. Natalie Bienstock (Moscow: Soviet Acade-my of Sciences, MEI), p. 170.

a Feodor Dostoevsky, Crime and Punishment,trans. Jessie Coulson, with Bechgrounds andSources and Essays in Criticism, ed, GeorgeGiblan (New York: Norton, 1964), P. 9.

o Dostoevsky, Crime and Punishment, p. 25,

A LITERARY ANALOG TO CONFLICT THEORIES SO

A more slovenly and degraded manner of lifecould hardly have been imagined, but It suitedRaskolnikov's present mood. He had resolutelywithdrawn from all human contacts, like atortoise retreating Into its shell, until the sighteven of the fate of the servant girl . . madehim shudder with revulsion,10

Upon awakening, he receives a letterfrom his mother, informing him of theupcoming marriage of his sister to aman he regards as a scoundrel. She alsotells him of her plans to visit him in thenear future. This event, which would al-ter his environment, "increases his ten-sion and brings him face to face withdecision ('Whatever happens, I mustdecide.1"11 In a state of anxiety ("Hisheart was beating fiercely and histhoughts were wildly agitated"),13 hegoes for a walk and observes a young,apparently drunk girl wandering aboutin confusion. She is being followed by adandy whom Raskolnikov suspects ofhaving immoral intentions. He arrangesfor her assistance and protection by apoliceman, and causes quite a commo-tion. As he departs, however, "an in-stantaneous revulsion of feeling"18 over-comes him and he reverses himself, tell-ing the policeman: "'Stop! What is itto you? Drop itl Let him amuse himself!'(he pointed at the gentleman). 'Whatbusiness is it of yours?' "14 Raskolnikov'ssudden change, of attitude toward thesituation may be attributed to his roleuncertainty as an extraordinary manand to: the ugliness of the entire event,which served to suddenly reinforce his"extraordinary man" theory; conse-quently, he decides to commit the mur-der.

This decision, like the other precedingit, is reversed shortly thereafter when,

10 Dostoevsky, pp. 25.26.11 Evnin, p. 171.12 Dostoevsky, p. 37.IS Dostoevsky, p. 47.14 Dostoevsky, p. 47.

In a park, Raskolnikov dreams of thebrutal beating of a horse. He is repulsedby the cruelty exhibited In his fantasy,and declares, "No, I shall not do it, Iwill not do iti"15 Dostoevsky describeshis mood in the next passage:

He stood up, looked round as if wonderinghow he tame to be there. . . He was pale,his eyes glittered, exhaustion filled every limb,but he had suddenly begun to breathe moreeasily. He felt, that he had thrown off theterrible burden that had weighed him downfor so long, and his heart was light and tran-quil. 'Lord,' he prayed, 'show me the way,that I may renounce this accursed . . fantasyof minet'Itt

But on his way home, he overhears aconversation in the Haymarket Square.The pawnbroker's sister, Lizaveta, is say-ing that she will not be home until 8;00pm. Raskolnikov, realizing this oppor-tunity, again suddenly reverses himselfand prepares to commit the crime,

II

At least three conflict models have lit-erary analogs in Crime and Punishment:Raskolnikov's plight illustrates whatDollard et al., have called the frustration-aggression hypothesis (and extensionsthereof);1? the extraordinary man theoryexemplifies the decision-making modelsproposed by Bernard,18 Rapoport,19 andMarkus and Tanter;20 and Dostoevsky'spresentation of the motives and needs ofRaskolnikov resembles the instinctual

15 Dostoevsky, p. 57.15 Dostoevsky, p. 57.It J. Dollard, L. W. Doob, N. E. Miller, 0. H.

Mowrer, and R. R. Sears, Frustration and dig.gression (New Haven: Yale Univ. Press, 1959),

IS Jessie Bernard, The Sociological Studyof Conflict," in International Sociological As.sociatIon, The Nature of Conflict: Studies onthe Sociological Aspects of International Ten.sions (Paris: UNESCO, 1957).

10 Anatol Rapoport, Fights, Gamer, and De-bates (Ann Arbor: Univ. of Michigan Press,1960).

20 Gregory B. Markus and Raymond Tanter,"A Conflict Model for Strategists and Man.agers," American Behavioral Scientist, 15 (1972),809.834.

60 SPEECH MONOGRAPHS

view of aggression which has been dis-cussed by such theorists as Ardreyai andLorenz as

The frustration-aggression hypothesis,as amended by Miller et al923 assertsthat "the occurrence of aggression al-ways presupposes the existence of frustration, and . frustration producesinstigations to a number of differenttypes of response, one of which is an in.stigation to some form of aggression." 24Extending this line of reasoning, New.comb formulated the autistic hostilityhypothesis:25

Briefly, [the thesis] is this: the likelihood thata persistently hostile attitude will develop varieswith the degree to which the perceived inter-personal relationship remains autistic, -Its pri-vacy maintained by some sort of barriers tocommunication. . If commUnicadon withothers is cut off, the initial framemirk respond.ble for the perception of hostility Is less likelyto be modified than if inter-personal give-and.take is continued.ge

In Crime and Punishment the inter.personal relationship in question existsbetween Raskolnikov and society, ratherthan with Alyona Ivanovna, the pawn-broker. His anxiety, resulting from theconstant interplay and exchange of per.sonalities (the intellectual vs. the com-passionate), as well as from the "slovenlyand degraded manner" in which helived, led to intense frustration. Theexistence of this potentially hostile at-titude is perpetuated by his inability toresolve his cognitive conflict, and is en-hanced by his determination to isolatehimself (in relation to Raskolnikov's

21 Robert Ardrey, African Genesis (New York:Atheneum, 1961).

22 Konrad Lorenz, On Aggression, trans. Mar-jorie Kerr Wilson (New York: Harcourt, 1966).

23 Neal E. Miller et al., ''The Frustration-Aggression Hypothesis," Psychological Review,48 (1941), 337.342.

24 Miller et al., M.25 Theodore M. Newcomb, "Autistic Hostility

and Social Reality," Human Relations, 1 (1947),

26 Newcomb, 69.

Isolation, Dostoevsky narrates, "Such be-haviour is found among monomaniacswhen they have concentrated all theirenergies on one pointm). Thus, thisself-imposed barrier to communicationprevented him from modifying or re.jecting his misanthropic theory. Dostoevsky indicates that Raskoinikov doesin fact renounce his theory in favor ofChristianity in the epilogue, and thathis decision, influenced primarily by hisrelationship with Sonia Marrneladova"Could not her beliefs become my be-liefs now? Her feelings, her aspirations,at least"28resulted from the breakdownof this communication barrier.

Raskolnikov's frustration and Isola.don is complemented by a motif whichrecurs throughout the novel: the motifof freshness. Prior to committing themurder, Raskolnikov visited the Peters.berg Islands, and the "greenery andfreshness" provided him with a tempo-rary respite from the "stuffiness, thejostling crowds, the bricks and the mor.tar" of the city. It was in this park thathe dreamt of the brutal beating of thehorse by the peasant Mikolka, the fifthperipateia of Part I and the last instancein which Raskolnikov is repulsed by theact he plans to commit, The freshnessmotif represents an escape for Raskolnikov, whereby he is able to free himselffrom the self-imposed isolation whichhad been the chamber of his torment,This self-imposed isolation, which hadcaused him so much anxiety, was initi-ated in conjunction with the sanctionsof his extraordinary man theory.

The theory had originally been con-ceived as a kind of cure for Raskolni-kov's unhappy situation. The extraordinary man theory obviated the necessityfor Raskolnikov to inhibit his actions,as he would answer to no one, and

21 Dostocvsky, Crime and Punishment, p. 26.25 Dostoevsky, p. 527.

A LITERARY ANALOG TO CONFLICT THEORIES 61

served only his own will. He was abovesocial and ethical constraints whichkept the ordinary man or "louse" inline. His ability to utter a "new word"enabled him to transgress social bar-tiers "in the name of better things."

113f It is necessary for one of them, for thefulfilment of his Ideas, to march over corpses,or wade through blood, then in my opinion hemay in all conscience authorize himself to wadethrough blood.'

The extraordinary man theory repre-sents another conflict model. The, au-tonomous being described is similar tothe decision-maker, and especially simi-lar to that group of strategists whichRapoport has labeled the "abstraction-ists'" and the "cool young men."30 Thesole criterion for conflict in this modelis whether an advantage is to be gained,regardless of moral, psychological, orother factors. An example of the de-cisionmaking model may be found informal theory, in which participantschoose among competing strategies inorder to maximize their gains and mini-mize their costs. Other factors, such asethics, values, or the psychological as-pects of the players, are normally ex-traneous to this approach. Markus andranter describe the value of this typeof decision-making model:

The value of such a model is the level of gen-erality it possesses due to its freedom fromhaving to take into account the Idioiyncrasiesof any actual situation. By excluding these per-turbations from its framework, the moglelachieves what is sometimes called 'elegance.%

These authors criticize game-theoreti-cal approaches on this very basis, how-ever, due to the inability of such frame-works to include individual psychologi-cal factors which interact in "real" situ-

Dostoevsky, p. 250.SOAnatol Rapoport, Strategy and Conscience

(New York: Harper, 1964), p. 177.at Markus and Tanter, 815.

ations. They support the use of the "sub-jectively expected utility" model (SEU),an approach which had previously beenused in analyzing games against na-ture.32 In adapting this framework toconflict theory, Markus and Tantercomment;

problem is transformed essentially intoan exercise in decisionmaking under risk. Theactor still does not know for sure what willhappen if he acts in a given way, but he canmake certain Probability statements about alternative outcomes. In the theory of risky de.dsion.making, the actor is assumed to maxi-mize his subjectively expected utility. . . .

The SEU maximization principle stems fromthe traditional mathematical notion of theexpected value of a game of chance. The ex.petted value of a bet is obtained by simplymultiplying the value of each possible outcome,or by the corresponding probability of occur-rence, pi, and then summing these productsacross all outcomes; Symbolically:88

EV = oips1=1

By comparing this theory with Raskol-nikov's extraordinary man concept, yetanother similarity between the twoframeworks may be demonstrated. Boththe SEU actor and the extraordinaryman embrace the notion of perceivedutility. As Raskolnikov explains to de-tective Porfiry Petrovich:

'[Hie may in all conscience authorize himselfto wade through bloodin proportion, however, to his idea and the degree of its impor-tancemark that. It is In that sense only thatI speak in my article of their right to commitcrime: as

Thus, the extraordinary man need onlyexamine the expected utility of his actin relation to his idea before its commis-

82 See, for example, Ward Edwards, "Be-havioral Decision Theory," Annual lieview ofPsychology, 12 (Palo Alto: Annual Reviews, Inc.,1961), 473-498; and R. Duncan Lute andHoward Raffia, Garnet and Decisions (NewYork: Wiley, 1957).

83 Markus and Tamer, 816.84 Dostoevsky, Crime and Punishment, 250.

62 SPEECH MONOGRAPHS

sign. Moreover, the furthering of hisnew idea is the only rationale he needsor has available to him. It is also note-worthy that Raskolnikov's final decisionwas influenced by his attainment of in.formation which signified a reductionof the probabte "risk of discovery."

Dostoevsky's motif of solitude or"aloneness ", relates to Raskolnikov's ex-traordinary man theory. This isolation ismental, not physical, as Dostoevsky ex-plains in his Notebooks for Crime andPunishment:

The truth of God and the law of nature taketheir own, and (Raskolnikov) finally feelsforced to give himself up, . . . in order to beOnce again part of humankind, even if it meansperishing in prison.30

There are several clues in the novelwhich imply that behavior is motivatedby an instinctual force. Philip Rahv statesin "Dostoevsky in Crime and Punish-ment" that the murder of the pawn.broker, although "intellectually ration-alized [is) inexplicable except in termsof an unconscious drive."36 The authordepicts Raskolnikov as a murderer insearch of a motive, simultaneously em-bracing and rejecting a host of reasonsfor his crime. As Rahv has observed:

The criminal himself is In his own fashionconstrained to take part in the work of de-tection . because he is soon lost in themare of his own motivation. Never quite cer-tain as to what It was exactly that Inducedhim to commit murder, he must continuallyspy on himself in a desperate effort to pene-trate his own psychology and attain the self-knowledge he needs if he is to assume responsi-bility for his absurd and hideous act.

Raskolnikov eventually forsook thepsychological quest for his motive: "Life

35 Fyodor Dostoevsky, The Notebooks forCrime and Punishment, ed, and trans. EdwardWasiolek (Chicago: Univ. of Chicago Press,1967), p. 172.

38 Philip Rahv, "Dostoevsky in Crime andPunishment," Partisan Review, 27 (1960), 425.

81 Rahv, 399.

had taken the place of logic and some-thing quite different must be workedout in his mind. "88. Dostoevsky attemptsto expose the futility of employing in-tellectual, dialectical methods in seek-ing to solve the mysteries of life. GeorgeGibian compares Dostoevky's rejectionof dialectics in favor of "life" to C. G.Jung's description of man's cognitiveprocesses.89 In a passage which Lavrinhas called "one of the best clues to Dos-toevsky the psychologist, "40 Jung charac-terizes misapplied intellect as:

thinking which Is a mere equation [italicsours), and from which nothing comes out butwhat we have put in.... [fileyond that there isa thinking in primordial Imagesin symbolswhich are older than historical man; 'which havebeen ingrained In him from earliest times, and,eternally living, outlasting all generations, stillmake up the groundwork of the humanpsyche.41

Jung's conception is related to Fretid'snotion of the "Primal Horde," and itseffects upon modern collectivities:

[The group appears to us as a revival of theprimal horde. just as primitive man survivespotentially in every individual, so the primalhorde may arise once more out of any randomcollection.42

Dostoevsky recognizes this revival inmodern man but interprets it in a morespiritual manner:

Not a single nation has been founded in prin.ciples of science and reason. There never hasbeen an example of it, except for a brief mo-ment of folly. . . . Nations are built up byanother force which sways and dominates them,the origin of which Is unknown and Inez-pltcabte. . . It is the spirit of life, as theScriptures call it, 'the river of living water,'

ss Dostoevsky, Crime and Punishment, p. 527.89 George Gibian, "Traditional Symbolism in

Crime and Punishment," PMLA, 70 (1955), 979996.

4o Janko Lavrin, Dostoevsky: A Study (NewYork: Macmillan, 1947), P. 53.

41 C. G. Jung, "The Stages of Life," ModernMan in Search of a Soul (New York: Harcourt,1934), pp. 129-150.

49 Freud, XVIII, p. 123.

A LITERARY ANALOG TO CONFLICT THEORIES 63

the drying up of which Is threatened by theA poatlypse.44

This passage, extracted from anotherDostoevsky novel, The Possessed, "couldserve as a basic text for Crime and Pun.ishment," according t Gibian.44 Thisspirit of life consists of not only evil andviolent tendencies rooted in the natureof men, but also includes the good andgentle aspects of man's soul. V. V. Zen-kovsky presents this dichotomy in a dis-cussion of Dostoevsky's anthropology:

Dostoyevsky exhibits not only the sin, corrup-tion, egoism, and, in general, the 'demonic'element in man with unprecedented force; beexhibits no less profoundly the impulses towardjustice and good in the human soul, the 'an-gelic' principle In man. The force and sigracanoe of Dostoyevsky's use of antinomies inphilosophical anthropology derives torn the factthat both of the opposites are presented intheir highest form.40

This view of man represents an an-thropological double contingency. Thatis, man's destiny is presented as a"thromise," Thromise, as defined byJohn Waite Bowers, refers to "circum-stances where a source controls bothnegative and positive sanctions for thetarget." 44 The target, in this context, isthe future of man, and the source is thatcombination of instinctive, unconscious,primordial motivations which comprisethe spirit of life.

III

In Crime and Punishment, Dostoev-sky presents the reader with three dis-tinct conditions of human existence:the self-willed misanthropic intellectual

4a Fyodor Dostoevsky, The Possessed (liar-mondsworth, Middlesex, England: PenguinBooks, Ltd., 1961), trans. David Magarshak.

44 Gibian, 979.45 V. V. Zenkovsky, "Dostoyevsky's Anthro-

pology," In A History of Russian Philosophy,trans. George L. Kline (London: Routledge andKegan Paul, 1953), 1, 418.

44 John Waite Bowers, "Beyond Threats andPromises," address delivered on May 16, 1973at Bowling Green State University, BowlingGreen, Ohio.

or extraordinary man, who is faithless(except in himself) and is set apart from"ordinary" people; the Christian, who Isdependent upon his faith and avoids at:tempting to logically comprehend themysteries of life; and those, like Raskol.nikov, who are torn between both ex-tremes,

While contemporary conflict theoriststend to take a monistic stance, attempt-ing to explain all human conflict interms of a single theoretical perspective,Dostoevsky operates from a fluid per-spective, demonstrating the variabilityof motivational factors based upon con-textual characteristics. In this manner,certain factors become salient dependingprimarily upon the parameters of theparticular situation. This conceptualiza-tion suggests that a synthesis of existingmodels may provide a more satisfactoryapproach to the study of human conflict.

Other implications of this essay sug-gest that science and art should beviewed as complementary investigativestrategies. In particular, the existence ofliterary analogs to conflict theories hasbeen demonstrated. That the analogspredate the conflict theories provides forthe complementary utilization of artand science via the application of propo-sitions derived from literary analysis totheory construction. Further applica-tions of literary analysis to the genera-tion of scientific theory in areas such aspsycholinguistics and intrapersonal andinterpersonal communication could ex-tend the scope of the interrelationship be-tween art and science. Andre Gide, in acomment on Dostoevsky's work which isapplicable to many authors, indicates therichness of literature as a source of ideas:

Had he been philosopher instead of novelist,he would certainly have attempted to bringhis ideas into line, whereby we should havelost the most precious of thenoi

4/ Andre Gide, Dostoevsky (London: Butlerand Tanner, Ltd., 1926), p. 48.

COMMUNICATION AND THE INDUCEMENT OFCOOPERATIVE BEHAVIOR IN CONFLICTS:

A CRITICAL REVIEW

DAVID W. JOHNSON

N order for a conflict to be managedconstructively, there must be ef-

fective and continued communicationamong the involved parties. Communi-cation is of basic importance in con-flicts; through communication partici-pants coordinate efforts at resolving theirdifferences, provide information concern-ing their position and intentions, venti-late feelings, reason together, bargain, ex-ercise influence, and expedite the devel-opment of settlements. The social impor-tance of conflict management and resolu-tion and the central role communicationplays in managing 'conflicts constructive-ly makes this area one of the most theo-retically significant in psychology.

DEFINITION OF CONFLICT

Deutsch broadly defines a conflict asa situation in which "incompatible ac-tivities occur"; an incompatible activityis any action that in some way makesanother action less probable or less effec-tive.1 More specifically, most studies ofconflict are conducted in the context ofa two-person, mixed-motive, incompleteinformation context. In a mixed-motiveconflict there is a cooperative interest inreaching an agreement since both partieswill be better off if an agreement isreached than if no agreement is reached,and a competitive interest for each partyto make the agreement as favorable tohimself as possible. In order for an agree-

Mr. Johnson is Professor of Psychology at theUniversity of Minnesota, Twin Cities.

I M. Deutsch, "Conflicts: Productive and Destructire," journal of Social Issues, 25 (1969),7.8.

SPEECH MONOGRAPHS, Vol. Il, March 1971

ment to L reached the cooperative in-terests of one or both parties must bestronger than the competitiVe interests.

A conflict in which there 4 incompleteinformation is a situation, in which theparties are not fully aware of thestrength of the competitive and coopera-tive forces on each other or the minimalpoint in the division of potential gainseach will agree to. There is a great dealof information dependence upon theother party in the incomplete inforMation conflict situation, as it is throughinteraction with the other that one'sexpectations concerning possible agree-ments are clarified. In the incompleteinformstion situation there is a basiccommunication dilemma: an agreementcan be most easily reached if both partiesare open and honest about their expecta-tions and work to ensure an equitableagreement. Yet if one party is hrne,stand the other is deceitful, the agreementwill be inequitable, in the direction ofbeing more favorable to the deceitfulparty than to the honest party. Eachperson has the choice to be honest ordeceitful in his communications about hisminimally acceptable agreement points,and each person must determine whichof these two alternatives the other haschosen.

DEFINITION OF COMMUNICATION

A major difficulty with the presentresearch on communication in conflictsituations is the lack of conceptualiza-tion for the concept "communication."In the research literature most investi-gators do not define communication con-

COOPFILITIPE BEIMPIOR CONFLICTS 65

ceptually. ,They seem to assume thateveryone knows what communication is,that its definition is widely known andagreed upon, ,and that no concertualdefinition is necessary. Nothing couldbe farther from the truth. The litera-ture in communication exhibits a multi-tude of definitions for the concept andlittle agreement about which definitionis most useful. Dance, for example, dida content analysis of 95 definitions ofcommunication from several diversefields; he derived 15 distinct conceptualcomponents of communication fromthose definitions. He notes that the va-riety of definitions has led differenttheorists and researchers in "differentand sometimes contradictory directions"and concludes that the concept is over-burdened and should be replaced witha "family of concepts."2

Adequate conceptualization is essen-tial for theoretical models to organizeand stimulate research. Concepts deter-mine the behavioral field observed,which affect the principles derived fromthe observations, which affect the hy-potheses, laws, and theories constructed.Without a conceptual definition of com-munication, or of various aspects of it,no theory of effective communication inconflict situations can be built. Becauseof the lack of conceptualization of com-munication, much of the research oncommunication in conflict situationstakes on the character of a blind manstumbling in the dark searching forsomething. to give him a frame of refer-ence concerning the ground he is tryingto explore.

Even the operational definitions ofcommunication used in the conflictstudies manifest wide differences, withsome researchers operationally definingcommunication as the passing of notes,

2 F. E. X. Dance, "The 'Concept' of Com-naunication," Journal *I Communication, 20(1970), 201-210.

others as the use of a telephone linkage,and others as face-to-face discussion.The nonverbal communication involvedin a person's appearance, gestures, facialcues, tone of voice, and so on, have mostoften been ignored. Thus while a writtenmessage stating that one intends to be-have cooperatively is considered a com-munication message, the cooperative be-havior which tacitly communicates co-operative intentions and a friendly ap-pearance and manner are not. Twoserious consequences ensue from suchvagaries in operationalizing communica-don. First, the operational definitionsused for communication have been inade-quate in clarifying the concept of com-munication. Merton notes that the clari-fication of concepts ordinarily enters intoempirical research in the shape of estab-lishing operational definitions of the vari-ables under consideration.8 The opera-tional definitions for an inexact, com-plex concept such as communicationshould explicate the concept with pre-cise specification in order to make futureresearch more theoretically significant.This has not been done in the conflictresearch. Second, the operationalizationsof communication have been unproduc-tive in advancing theory. The purpose ofan operational definition is to serve asan index of a concept which is related toother concepts within a theory. Opera-tional definitions serve the purpose ofdefining or partially defining concepts interms of observable data in order totest the empirical validity of a theory.The lack of conceptualization of com-munication and the inadequacy of theoperational definitions used combine tomake most of the research in this areaunproductive in advancing theory. Thusthe operational definitions conflict re-

SR. X. Merton, 'The Bearing of EmpiricalResearch on Sociological Theory," in SocialTheory and Social Structure, rev. and enl. ed.(New York: Free Press, 1957).

68 SPEECH MONO 0 RA PBS

searchers have used are often inconsis-tent with most conceptual definitions ofcommunication, have increased the con-fusion concerning the conceptual defini-tion of communication, and are a majorMSC of the lack of theoretical produc-tivity in the area, The lack of comp-tual definitions of communication andthe confusion caused by inappropriateoperationalizations of communicationmake it imperative that some clarity bebrought to this area of conflict research.

In summarizing several of the defini-tions of communication Johnson writesthat because (if there Is pereeptual en-gagement) we continuously affect oneanother (altering perceptions, disposi-tions and expectations), interpersonalcommunication can be defined verybroadly as any behavior, verbal or non-verbal, that is perceived by another persons Tnterpersonal communication, how-ever, is more commonly and specificallydefined as a person sending a messageto a recipient(s) with a conscious intentto affect the latter's behavior. Electivecommunication can then be defined asexisting between two persons when thereceiver interprets the sender's messagein the same way the sender intended it.This definition of communication doesnot mean there is always a temporal se-quence of events whereby a personthinks up a message, sends it, and some-one else receives it. Communicationamong individuals is a process in whicheveryone receives, sends, thinks, inter-prets, and so on, all simultaneously andthere is no beginning or end. Almost allof the research reviewed is experimentaland, therefore, has created a temporalsequence of sending and receiving mes-sages as part of proving causation. Smithnotes that such research is contradictory

t D. W. Johnson, Reaching Olaf interper-sonal Effectiveness and Self - Actualization (Engle-wood Cliffs, N. _T : Prentice-Hall. 1912); andHumanistic Social The tny and Re.search (Philadelphia: Lippincott, IWO).

to the process view of communicativeand, therefore,' may be invalid.8 Finallyit should be noted that communicationinvolves the transmission among individuals of symbols to which certainmeanings are attached, These symbolscan be either verbal or. nonverbal. Theexchange of ideas and experiences ameegindividuals is poillible only when bothhave adopted the same conventions forrelating a particular graphic, nonverbal,or spoken symbol to a particular concep-tual experience.

Research on communication in conflictsituations has by and large ignored thetheoretical literature in communication.This results in misdirected research whichis irrelevant to current communicationtheory, =repeats work that has *headybeen done, and results in dubious ai,Wrsumptions being made about the com-munication process, all of which mayinvalidate much of the research whichhas been conducted. In addition, thereis frequently an unbridged gap betweenthe results found by researchers and theapplication of the findings to.other con-flict situations. The social importance ofthis area makes imperative researchto validate communication procedureswhich can be used in actual interpersonaland inter-group conflicts. In concep-tualizing the communication process inconflict situations, therefore, emphasisshould be placed upon utilizing the com-munication literature and selecting var-iables which can be easily implementedand operationalized in "practical" situ-ations.

In order to begin building a theoreticalmodel of communication effectivenessin conflict situations the concept "com-munication" has to be subdivided. Thecentral research task is to establish theconditions under which certain types of

5 D. H. Smith, "Everyone Talks About Proc-eu But No One Does Anything About It," Uni-versity of Minnesota, 1972 (Mimeographed.)

COOPEnATIPE 11014PIOR CONDUCTS 67

messages sent through certain types ofchannels in the context of certain situ-Ational variables will be received in sucha way as to influence the receiver's de-cision to respond cooperatively. A chan-nel is the means of conveying a messageto another person; technically, thesoundwaves of the voice and the light-waves involved in seeing are the channelsfor much of the communication whichtakes place among individuals. A menageis defined as any verbal or nonverbalstimulus that one person transmits toanother; it refers to some informationabout a referent in a symbolic repre-sentation,

DEPENDENT VARIABLES

The dependent variable most oftenused in the studies on communication inconflict situations is the quantity of co-operative behavior. It is doubtful wheth-er this is the most appropriate dependentvariable to use. There is a large differ-ence between accurately receiving thecontent of a message, making attributionsconcerning the intentions of the sender,making a decision to respond in either acooperative or a competitive way, and ac-tually responding with cooperative beh av-ior. If, for example, a subject responds to acooperative message with competitive be-havior it is not clear whether he mis-construed or misperceived the message,distrusted the sender's intentions, ordecided to exploit the person's coopera-tive intentions for incentive, personality,or other reasons. The dilemma involvedin using messages in honest or deceitfulways may often influence a subject todisbelieve an ostensibly cooperative ma-sage. What is needed is a series of studieswhich focus upon the dependent vari-ables necessary to illuminate the proc-esses by which message characteristicsaffect the induction of cooperation.

Most of the games used in conflict ex-

periments provide more options to a play-er faced with a cooperative opponent thanto one faced with a competitive opponent.When an opponent is cooperativelyoriented, the individual can gain whetherhe cooperates or competes, With a com-petitive opponent, only a competitiveresponse will minimize the subject'slosses. Thus the decision to respond co-operatively to a cooperative messagemay be complex. Ignoring this processprovides only minimal understanding ofbehavior in conflict situations.

REVIEW OF PRESENT KNOWLEDGE

Given the conceptual problems withthe research dealing with the use of com-munication to induce cooperative be-havior in conflict situations, the taskremains to describe whit research hasbeen done and to assess what you havewhen you put it all together. In the fol-lowing sections the available researchis reviewed and grouped into three cate-gories: (1) studies allowing game be-havior as the only means of communica-tion, (2) studies allowing communicationthrough game behavior and worded mes-sages, and (3) studies allowing com-munication through game behavior,worded messages, and nonverbal mes-sages.

STUDIES ALLOWING GAME BEHAVIOR

AS THE ONLY MEANS OF COMMUNICATION

There are numerous studies which al-low as the only channel for transmittingmessages the choices made in bargaininggames such as the prisoner's dilemma(PD) game. In these studies differenttypes of strategies are used to inducecooperative behavior on the part of thesubject. A /oaten may be defined as apreplanned program of choices (includ-ing programs which have elements ofrandomness or probabilistic responding)in a game situation, regardless of whether

68 SPEECH MONOGRAPHS

it is implemented by a subject, by a con-federate of the experimenter acting asthe only player, by the experimenterhimself, or by a computer, Strategiesthen become a message indicatingthrough behavior the resultant of theinternal and external forces on the senderto cooperate or compete. Needless to say,this is a severely limited communicationsituation. It is assumed that all sub-jects attend to the pattern of choicestheir opponent makes and that the cor-respondence of meanings attached tochoice patterns is high. There is no op-portunity for simultaneous redundancyto increase the accuracy of the communi-cation. It is especially difficult to ascer-tain intentions in a two-choice game, asa particular choice can represent an in-dividualistic strategy, a mere responseto the other's previous choice, an invita-tion to adopt a certain strategy, a pun-ishment, or a signal of aggressiveness orcooperativeness. The absence of verbal,auditory, and visual channels of corn-munication may promote atypical be-havior on the part of some subjects.

Although the studies reviewed belowdemonstrate that certain behavioralstrategies are more effective in inducingcooperative behavior than others, they donot demonstrate why. There are a vari-ety of post hoc explanations presentedto explain the results, but these explana-tions have not been adequately tested.Most of these studies have not testedtheoretical hypotheses; they have merelytested hypothesized relationships be-tween various strategies and the numberof cooperative choices subjects make inthe game. Theoretical explanations ofwhy the strategies should affect the sub-ject's game behavior and the measuringof intervening variables are noticeablylacking. Consequently, much of the re-search in this area is of little or no sig-nificance.

The studies reviewed will be organized

in the following categories: (1) non-contingent vs. contingent strategies, (2)abrupt changes in noncontingent strate-gies, and (3) multi-choice games.

Contingent vs. Noncontingent Strategies

Many of the studies of behavior inbargaining games can be classified astesting the efficiency of contingent ornoncontingent strategies in inducing co-operative behavior on the part of sub-jects. Perhaps the major difference be-tween contingent and noncontingentstrategies from a communication point ofview is that the contingent strategy isresponsive to the subject's behaviorwhile the noncontingent strategy is un-responsive to the subject's behavior. Ingeneral, the research indicates that avariety of noncontingent strategies arenot highly effective in inducing coopertetion.6

There are two formats for studyingthe effects of contingent, matching stra-tegies in bargaining games. The firstinvolves sequential play in which, onany one trial, the subject makes hischoice and then the confederate chrose,s.Matching takes place on the same trialunder these conditions. In the sequential

ay. B. Bixenstine, H. M. Potash, and K. V.Wilson, "Effects of Level of Cooperative Choiceby the Other Player on Chokes in a Prisoner'sDilemma Game: Part I," journal of Abnormaland Social Psychology, 66 (1963), -313; V. E.BbcenstIne and K. V. Wilson. "Effects of Levelof Cooperative Choke by the Other Player onChoices in a Prtsoner's Dilemma Game: PartIL" Journal of Abnormal and Social Psychology,67 (1963), 139.147; A, A. Harrison and C. -G.McClintock, "Previous Experience Within theDad and Cooperative Game Behavior," Journalof Personality and Soda! Psychology, 1 (1965),6 1-675; C. G. McClintock, A. A. Harrison, S.Strand, and P. Gallo, "Internationalism-lsola-tionism, Strategy of the Other Player, and Two.Person Game -Behavior," journal of Abnormaland Social Psychology, 67 (1963), 631.636; W.Wilson, "Cooperation and the Cooperativenessof the Other Player," Journal of Conflict Reso.lution, 13 (1969), 110-1E7; and C. D. McKeown,J. P. Gahagan, and J. T. Tedeschi, "The Effectof Prior Power Strategy on Behavior After aShift of Power," Journal of Experimental Re.search in Personality, 2 (1967), 226.233.

COOPERATIVE BEHAVIOR CONFLICTS 69

play situation, a same trial matchingstrategy in which the subject choosesfirst has been found to tproduce muchgreater cooperation than neither a 100%cooperative or a 0% cooperative stra-tegy° and either a 90% cooperative or a10% cooperative strategy .°

The second format for studying con-tingent, matching strategies is a simul-taneous choice in which matching alwaysfollows the choice made by the subjectby one trial. The communication problemwith the simultaneous choice situation isthat a subject does not know the extentto which his own choice influenced thepartner's next choice or whether theother player's intentions are reflected inhis choice. In other words, the externaland internal "causes" of the behavior arenot clearly separated. With simultaneousplay, matching -strategies regularly pro-duce significantly more cooperation thandoes 0% cooperations but not necessarilymore than does 100% cooperation.10 Sev-eral studies have demonstrated that amatching strategy produces significantlymore concurrent cooperation than doesa free play condition.11 A study by. Ser-mat also demonstrated the effectiveness

/ L. Solomon, "The Influence of Some Typesof Power Relationships and Game StrategiesUpon the Development of Interpersonal Trust,"Journal of Abnormal and Social Psychology, 61(1960), 223.230.

R. H. Whitworth and W. C. Lucker,"Effective Manipulation of Cooperation withCollege and Culturally Disadvantaged Popu-lations," Proceedings, 77th Annual Convention,American Psychological Association, 4 (1969),305406.

0C. M. Crumbaugh and G. W. Evans, "Pre.sentation Format, Other Person Strategies, andCooperative Behavior in the Prisoner's Dam.ma," Psychological Reports, 20 (1967), 895.902;and W. Wilson, "Cooperation and Cooperative-ness."

10S. Oskamp and D. Perlman, ''Factors Affecting, Cooperation in a Prisoner's DilemmaCame,' of Conflict Resolution, 9 (1965),359-974; and W. Wilson, "Cooperation and Co.operativeness."

11 L. Downing et al, "Profit vs. Social Mo-tives In the Prisoner's Dilemma Game," paperpresented at the Midwestern Psychological As.Iodation convention, Chicago, May 2, 1968; S.Oskamp, "Effects of Programmed Initial Strat-egies in a Prisoner's Dilemma Game," Psy-

of a matching strategy in inducing co-operative behavior, and a matchingstrat-egy was shown by Crurnbaugh andEvans to produce significantly greater co-operation than a noncontigent strat-egy having exactly the same level ofcooperation." Finally, in a study in-volving several variations of matchingstrategies, Eixenstine and Gaebeleinfound that a strategy which is slow toreciprocate cooperative behavior andslow to reciprocate competitive behaviorwas most effective in inducing coopera-tive behavior from subjects."

Abrupt Changes in Strategy

Although no theoretical rationale hasbeen presented to predict the findings,several studies have demonstrated thatabrupt changes in game strategy affectsubject's behavior. A sudden shift fromhigh competition to high cooperationproduces more concurrent cooperationthan does high cooperation throughout.14A sudden shift from high competition toa matching strategy has been demon-strated to induce more concurrent co-operation than does a matching strategythroughout or a matching strategy pre-

chcmomic Science, 19 (1970), 195.196; W. Wilson,"Reciprocation and Other Techniques for In-ducing, Cooperation in the Prisoner's DilemmaGame; Journal of Conflict Resolution, 15(1971), 167.195; M. Pilisuk, P. Skolnick, and E.Overstreet, "Predicting Cooperation From theTwo Sexes in a Conflict Simulation," Journalof Personality and Social Prhology, 10 (1968),35.43; and M. Pilisuk and P. Skolnick, "Induc-ing Trust: A Test of the Osgood Proposal,"journal of Personality and Social Psychology, 8(1968), 121-133.

12 V. Sertnat, "The Effect of an Initial Co.operative or Competitive Treatment Upon aSubject's Response to Conditional Cooperation,"Behavioral Science, 12 (1967), 301.313; andCrumbaugh and Evans.

IS V. E. Bixenstine and I. W. Caebelein,"Strategies of 'Real' Opponents in Eliciting.Co.operative Choice in a Prisoner's DilemmaGame," Journal of Conflict Resolution, 15(1971), 157-166.

14 A. Scodel, "Induced Collaboration in SomeNonZero-Sum Games," Journal of Conflict Res.olution, 6 (1962), 335.340; and Amnon Rapoportand A. Mowshowitt, "Experimental Studies ofStochastic Models for the Prisoner's Diiemma,"Behavioral Science, 11 (1966), 444-458.

;0 SPEECH MONOORAPHS

ceded by a high level of cooperation."An abrupt shift from a high percentageof competitive choices to a high percent-age of cooperative choices induces morecooperative behavior than does anabrupt shift from a high percentage ofcooperative choices to a high percentageof competitive choices," although Ko-morita and Mechling found contradic-tory evidence, and Swingle and Coadyfound no significant differences amongconditions which included shifts towardmore or less cooperation, although varia-bility of the subjects' choices did in-crease." Teger, in an attempt to recon-cile these diverse findings, conducted anexperiment in which he varied the sizeof the hostile act that either did'or didnot follow a series of cooperative re-sponses by an opponent. The results in-dicated that a hostile act which is pre-ceded by cooperation appears more hos-tile and evokes greater retaliation thanwhen the hostile act is not preceded bycooperation. In addition, the size of thehostile act following cooperation makesa difference; the retaliation was de-creased somewhat when the hostile actwas small."

Swingle found that a subject's reactionto a shift in strategy by an opponent is

10 T. Hanford and L. Solomon, "'ReformedSinner' and 'Lapsed Saint' Strategies in thePrisoner's Dilemma Game," Journal of ConflictResolution, 11 0967), 104.109; and W Wilson,"Reciprocation."

laBixenstine and Wilson, "Cooperative ChoicePart II"; P. G. Swingle, "Effects of PriorExposure to Cooperative or Competitive Treat-ment Upon Subject's Responding in the Privoner's Dilemma,' Journal of Personality andSocial Psychology, 10 (1968), 44.52; and Oskamp.

it S. S. Romorita and J. Mechting, "Betrayaland Reconciliation in a Two-Person Came,"Journal of Personality and Social Psychology, 6(1967), 549-353; and P. G. Swingle and H. Coady,'Effects of the Partner's Abrupt Strategy ChangeUpon Subject's Responding in the

Strategy

Dilemma," Journal of Personality and SocialPsychology, 5 (1967), 357-363.

le A. I. Teger, 'The Effect of Early Coopera-tion on the Escalation of Conflict," Journal ofExperimental Social Psychology, 6 (1970), 187-204.

a function of his initial competitiveness.Highly cooperative subjects retaliatedagainst the opponent's defection immedi-ately and severely while highly competi-tive subject* showed a tendency to be-come more cooperative following theopponent's shift to increased competi-tion." McClintock, Gallo, and Harrisonfound that subjects who were "interna-tionalists" were responsive to variationsin an opponent's strategy but political"isolationists" were not." Swingle andGillis found that subjects were influencedmore by abrupt changes in strategy byliked others than by disliked others."

There. are several post hoc explana,tions for these findings, It has been sug-gested that initial competitiveness com-municates a willingness to be competitivewhich deters future competitive behavioron the part of subjects after a switchtowards cooperation has taken place.From a learning point of view the initialperiod of competition, which generallyleads to mutual competition, may givethe subject a chance to learn that thecompetitive behavior is a very possiblebut punishing outcome. Early coopera-tive behavior, on the other hand, givesthe subject a chance to learn that com-petitive behavior can be highly reward-ing, a lesson he must unlearn if mutualcooperation is ever to be achieved. Final-ly, Kelley notes that there is a tendencyto attribute to oneself those actions ofanother that are consistent with one'sown interests. He writes that a possibleexplanation for the findings that an in-crease in cooperation results in morecooperation than does consistently high

10 SwIngle.20C. G. McClintock, P. Gallo, and A. A.

Harrison, "Some Effects of Variations in OtherStrategy Upon Game Behavior," Journal ofPersonality and Social Psychology, 1 (1965), 319.326,

It P. G. Swingle and J S. Gillis, "Effects ofthe Emotional Relationship Between Protagon-ists in the Prisoner's Dilemma," Journal ofPersonality and Social Psychology, 8 (1968), 160.165.

COOPER4TWE BEIMP1OR CONFLICTS 71

cooperative behavior may be that thesubjects believe that their game be-havior, not the intentions of the other,"caused" much of this change."

It is apparent that no theoreticalframework has been developed and testedto explain the above findings. Althoughthe subjects are faced with abruptchanges in their, opponent's behavior,which may signal a variety of intentions,there is no readily identifiable explana-tion for what the content of the messagebeing communicated actually is.

Multi - Choice Games

The results of the research using multi-choice games, which allow a more preciseand unambiguous communication of in-tentions, in general show a higher re-ciprocity of cooperation than is normallyfound in the two-choice situations. Oneset of studies has utilized an expandedPD game allowing each player a choiceof six moves instead of the usual two;this permits the communication of de-grees of cooperativeness and competi-tiveness, In addition, the opportunity touse a signaling device to communicateone's planned behavior before actuallymaking a choice in the game has beenvaried. A matching strategy and a con-ciliatory strategy, in which the confeder-ate was slightly more cooperative thanwas the subject, induced more coopera-tive behavior than was present in a con-trol group consisting of pairs of subjectsactually playing against each other."When the signaling device was used withintegrity, increases in cooperative be-havior tended to follow; there is, how-ever, a tendency to make. deceptive useof the signaling of planned behaviorwhich results in a decrease of trust andan increase in competitive behavior."

112H. H. Kelley, "Attribution in Social Inter-action" (General Learning Press, 1971).

2$ Filial* and Skolnick.24 Pilisuk and Skolnick; and M. Pilisuk, J. A.

Pilisuk, Potter, Rapoport, and Winterfound that in those pairs of subjectswhere both players had .taken a substan-tial unilateral initiative towards coop-erative behavior at some point early inthe course of play, a pro ;nosis for mutualcooperation was good.26

Wilson and Bixenstine added a thirdchoice in the PD game which allowed aplayer to communicate a desire to co-operate without suffering excessive lossto himself or excessive gain to his op-ponent." Their data do not show anyappreciable increase in the number ofjoint cooperative responses as a resultof having" this alternative. Komorlta;Sheposh, and Braver found that the useof a third choice in the PD game to com-municate "I'll cooperate if you will andwe will both profit equally, but if you donot cooperate, I will punish you," in-duced more cooperative behavior thandid the use of a third choice to commti.rate "I have the advantage and I inteMto use it to the utmost," or "I will notuse my power over you.""

Deutsch, Epstein, Canavan, and Gum-pert used a more complicated bargaininggame in which a variety of choices wereavailable to players. Their experimentdemonstrates that neither a punitive nora rewarding response to noncooperativebehavior is more effective in elicitingmutually rewarding cooperation. Re-warding noncooperative behavior leads

Winter, R. Chapman, and N. Haas, "Honesty,Deceit, and Timing in the Display of Inten-tions," Behavioral Sdence, 12 (1967), 205-215.

25 Pilisuk, P. Potter, Anatol Rapoport,and J. A. Winter, "War Hawks and PeaceDoves: Alternate Resolutions of ExperimentalConflicts," Journal of Conflict Resolution, 9(1965), 491-508.

24 K. V. Wilson and V. E. Bixenstine, "Effectsof a Third Choice on Behavior in a Prisoner'sDilemma Game," Nebraska Psychiatric Intl-Lute and Kent State University, 1962. (Mimeo-graphed.)

27S. S. Komorita, J. P. Sheposh, and S. L.Braver, "Power, the Use of Power, and Co-operative Choice in a Two-Person Game,"Journal of Personality and Social Psychology, 8(1968), 134-142.

72 SPEECH MONOGRAPHS

to exploitation unless it has been pre-ceded by a convincing display of aggres-sive potential. Punishing noncooperativebehavior makes it difficult for a subjectto perceive cooperative intent, for thethreatening and aggressive nature of thebehavior designed to deter noncoopera-tion interferes with that perception. Astrategy which does not reciprocate hos-tility but nevertheless does not allow itto be rewarding seems to be effective ineliciting cooperative behavior so long ask is also generously responsive to theother's cooperative behavior."

Summary

The findings of these studies can beexplained post hoc as supporting thepropositions that a message which is re-sponsive to the other person's messageswill have more impact than will a non-responsive message, serial redundancy ishelpful in inducing cooperative behav-ior, dramatic switches in behavior mayincrease the impact and effectiveness of amessage, and messages which unambig-uously communicate intentions are moreeffective than messages which are am-biguous about intentions.

STUDIES ALLOWING COMMUNICATION

THROUGH GAME BEHAVIOR AND

WORDED MESSAGES

A number of studies have operation-alized communication by providing thedual channels of game behavior andworded messages. The worded messageshave consisted of prepared notes a sub-ject could send, the opportunity to writevotes of any sort to the opponent, talk-ing to the opponent on a telephone link-age which distorted voice tone and in-flection, or using experimental apparatusin ways which could indicate one of a

28 M. Deutsch, Y. Epstein, D. Canavan, andP. Gumpert, "Strategies of Inducing Cooperation: An Experimental Study," journal ofConflict Resolution, 11 (1967), 345-360.

number of possible worded messages. Thefindings of the studies can be discussedand summarized under three broadareas: (1) the presence vs. absence ofcommunication channels, (2) the con-tent of the messages sent, and (3) situ-ational variables.

Pretence vs. Absence ofCommunication Channels

There is a great deal of evidence thatthe presence of a channel in whichworded messages can be sent and ex-changed, contrasted with the absence ofsuch a channel, clearly Ancreases theamount of cooperative behavior foundin bargaining games." Swensson, how-ever, conducted a study which did notfind significantly more cooperative be-havior in conditions allowing for theexchange of worded messages contrastedwith a condition allowing only for com-munication by game behavior.80 Shure,

29 M. Deutsch, "Trust and Suspicion," Journalof Conflict Resolution, 2 (1958), 265.279; "TheEffect of Motivational Orientation Upon Trustand Suspicion," Human Relations, 13 (1960),123.139; j, L, Loomis, "Communigtion, theDevelopment of Trust, and Cooperative Be-havior," Human Relations, 12 (1959), 305.315;M. Deutsch and R. M. Krauss, 'Studies of In-terpersonal Bargaining," Journal of ConflictResolution, 6 (1962), 52.76; G. Evans, "Effectof Unilateral Promise and Value of RewardsUpon Cooperation and Trust," Journal ofAbnormal and Sodal Psychology, 69 (1964), 587-590; G. H. Shure, It. J. Meeker, and E. A.Ilansford, "The Effectiveness of Pacifist Strate-gies in Bargaining Games," journal of ConflictResolution, 9 (1965), 106.117; It. Radlow andM. P. Weidner, "Unenforced Commitments in'Cooperative' and 'Noncooperative' Non-Con-stant-Sum Games," Journal of Conflict Resolu-tion, 10 (1966) 497-505; V. Daniels, "Communi-cation, Incentive, and Structural Variables inInterpersonal Exchange and Negotiation," Jour-nal of Experimental Social Psychology, 3 (1967),47.74; K. W. Terhune, "Motives, Situation,and Interpersonal Conflict Within Prisoner'sDilemma," Journal of Personality and SocialPsychology, Monograph Supplement, 8 (1968);and P. G. Swingle and A. Sand, "Communica-tion In Non-Zero-Sum Games," Journal of Per-sonality and Social Psychology, 23 (1972), 54-63.

30 R. G. Swensson, "Cooperation in the Pris-oner's Dilemma Game I: The Effects of Asymmetric Payoff Information and Explicit Com-munication," Behavioral Science, 12 (1967), 314-322.

COOPERATIVE BEHAVIOR CONFLICTS 73

Meeker, and Hansford found evidencethat white worded messages producedmore cooperative behavior in a limitednumber of subjects, "for many otherstheir resolve to dominate was strength-ened, or at least rationalized, by attribu-ting trickery" to t ",he message sender.There is evidence that, white "anticipatedopportunity" to exchange worded mes-sages "enhanced initial cooperativeness inthe PD game,"32 bargainers who have thechannel available "but do not make useof it to communicate either an appeal forcooperation or equity tend to be aboutas successful" in increasing the level ofcooperative outcomes as pairs of bar-gainers who have the opportunity tocommunicate only through their gamebehavior." The overall evidence indi-cates that the use of two channels (i.e.,behavior in the bargaining game andworded messages) is clearly superior tothe use of only the bargaining game be-havior as a message to induce coopera-tive behavior. These findings support thenotions that the greater the number ofchannels used the more effective a mes-sage will, be in inducing cooperative be-havior and that the greater the simul-taneous and serial redundancy of mes-sages aimed at inducing cooperative be-havior the more effective they will be.

Message Content

As a message increasingly allows forexpression of intentions, expectations,conditions of retaliation and reconcilia-tion, and negotiation it will be more suc-cessful in inducing cooperative behav-ior." Terhune found that the occur-rence of messages which specifically re-duced the ambiguity of intentions andexpectations was correlated with amount

3i Shute et al., 114.32 Swingle and Santi, 54.33 Swingle and Santi, 61; and Deutsch and

Krauss, "Studies of Interpersonal Bargaining."84 Loomis; Evans; Shure et al.; Radlow and

Weidner; Daniels; and Swingle and Santi.

of cooperative behavior.85 Krauss andDeutsch conclude that messages in-tended to be fair and aimed at increas-,.ing the possibilities of cooperation aredesirable in inducing cooperative be-havior, a conclusion supported bySwingle and Santi." Messages whichemphasize reciprocity of choice, the de-sirability of cooperative choices, orthreatened penalties for noncooperativechoices are all effective in inducing co-operative behavior." All these studiessupport the notion that the more infor-mation contained within messages in-dicating the sender's cooperative inten-tions and expectations and the ways inwhich cooperative behavior can be co-ordinated, the more effective the mes-sages will be in inducing cooperativebehavior.

Situational Variable;

A variety of situational variables havebeen found to affect the level of co-operative behavior within the studieswhich have allowed communicationthrough game behavior and worded mes-sages. There is evidence that the greaterthe competitiveness of the situation, theless communication channels will be usedor the more likely they will be used todeceive, threaten, or insult the oppon-ent." Krauss and Deutsch found thatmessages are more effective in inducingcooperative behavior if they are intro-duced after the bargainers have experi-enced the destructive effects of mutual

33 Terhune.30 R. M. Krauss and M. Deutsch, "Communi-

cation in Interpersonal Bargaining," Journal ofPersonality and Social Psychology, 4 (1966), 572 -577; and Swingle and Santi.

31 Radlow and Weidner.88 Deutsch, "Trust and Suspldon" (1958);

Deutsch and Krauss, 'Studies of InterpersonalBargaining"; Pilisuk et al., "Honesty, Deceit,and Timing"; Terhune; and D. Wallace andP. Rothaus, "Communication, Group Loyalty,and Trust in PD Game," Journal of ConflictResolution, 13 (1969), 370.380.

74 SPEECH MONOGRAPHS

competition." Evans found that when apowerful third party (i.e., the experi-menter) indicated that he would penal-ize anyone who used a communicationchannel to deceive the opponent, co-operation was enhanced;4° Krauss andDeutsch found that when the experi-menter set norms for how the commun-ication channel should be used cooper-ation was facilitated." Under competi-tive conditions, Shure, Meeker, andHansford found that subjects thoughtthe message sender was trying to de-ceive, trick, embarrass them or makethem feel guilty." There ;5 some evi-dence that task complexity affects com-munication, effectiveness; for very simpletasks which need to be completed in ashort period of time the use of com-munication channels to exchange wordedmessages may interfere with task ac-complishment." Swensson and Terhunepresent some evidence that the sendingof a message indicating that one willbehave cooperatively may function asa commitment to do so, which increasescooperative behavior." Making one'sworded and behavioral messages con-gruent increases the probability of suc-cessfully inducing cooperative behav-ior." A communicated commitment tobe cooperative will carry more weightwhen the person has behaved in trust-worthy ways in the past."

31 Krauss and Deutsch, "Communication inInterpersonal Bargaining."

40 Evans.41 Krauss and Deutsch, " Communication in

Interpersonal Bargaining."44 Shure et al.43C. B. McConville and J. R. Hemphill,

"Some Effects of Communication Restraints onProblemSolving Behavior," Journal of SocialPsychology, 69 (1966), 265-276.

44 Swensson; and Terhune.46 5. P. Cahagan and 5. T. Tedeschi, "Strategy

and the Credibility, of Promises in the Pris-oner's Dilemma Game," Journal of ConflictResolution, 12 (1968), 224.234; and . Horaland I. T. Tedeschi, "Effects of Credibility andMagnitude of Punishment on Compliance toThreats," Journal of Personality and SocialPsychology, 12 (1969), 164.169.

Cahagan and Tedeschl.

STUDIES ALLOWING COMMUNICATION

THROUGH GAME BEHAVIOR, WORDED

MESSAGES, AND NONVERBAL MESSAGES

As different channels for the se tidingof messages become available, not onlyare a larger number of messages madeavailable for simultaneous. transmission,but possible different kinds of meaningsare communicated. By being able to seeeach other, for example, the subjectsare provided with a set of nonverbalmessages that provide a context in whichthe explicit verbal messages becomemore reliable. In a normal two-personconversation the verbal componentshave been estimated to carry less than35 percent of the social meaning ofthe situation" and, therefore, the pres-ence of nonverbal messages is importantin inducing cooperative behavior. Wecommunicate by our manner of dress,physique, postures body tension, facial ex-pression, degree of eye contact, hand andbody movements, tone of voice, con-tinuities in speech such as rate, dura-tion, inflections, nonfluencies, and pauses,spatial distance, and touch, as well asby words, By comparison with verballanguage, however, nonverbal messagesare limited, ambiguous, and difficult tointerpret accurately." Usually nonverbalmessages communicate feelings, likings,and preferences in ways which rein-force or contradict verbal messages.

Comnwnication Channels

The studies of communication chan-nels in situations allowing game behav-ior, worded messages, and nonverbalmessages can be divided into studiesfocusing upon the alternation of differentchannels and studies focusing upon thesimultaneous use and comparison of

47 Ray L. Birdwhistell, Kinesics and Context..Essays on Body Motion Communication (Phila.delphia: Univ. of Pennsylvania Press, 1970),157.158.

48 Johnson, Humanistic Social Psychology.

COOPERATIVE BEHAVIOR CONFLICTS 76

different channels. There have been aseries of studies in which the subjectsplay a bargaining game allowing onlythe game behavior as communication,then have an intermission in which thesubjects could or could not communicatewith each other face-to-face, and thenfurther play the bargaining game withgame behavior as the only means of com-munication. The results of these studiesindicate that while very short timeperiods of unstructured, face-to-face com-munication do not increase cooperativebehavior in a bargaining situation,"longer periods of unstructured, face-to-face communication do increase cooper-ative behavior."

There is evidence that the simul-taneous use of all verbal and nonverbalcommunication channels is characteristicof successful conflict management andresolution. Families which successfullymanage conflict, for example, exchange in-formation through verbal and nonverbalchannels much more frequently than dofamilies which are unsuccessful in man-aging conflict." There are a variety ofstudies, furthermore, which deal withcomparisons among behavioral, visual,auditory, and total nonverbal and verbalchannels. There is evidence that the corn-

49 A. Scodel, J. S. Minas, P. Ratoosh, and M.Lipetz, "Some Descriptive Aspects of TwoPerson NonZeroSum Gamer, " lournal of Con.flier Resolution, 3 (1959), 114.119.

60 Anatol Rapoport, A. Chammah, J. Dwycr.and J. Gyr, "Three-Person Non-Zero.Sum Non.neptiable Games, Behavioral Science., 1 (1962),3848; V. E. Bixenstine, C. A. Levitt, and K. V.Wilson, "Conaboration Among Six Persons ina Prisoner's Dilemma Game," journal of Con-flict Resolution, 10 (1966), 488.496; and V. E.BIxenstIne and J. Douglas. "Effect of Psycho-pathology on Group Consensus and CooperativeChoice in a Six..Person Game, Journal of Personality and Social Psychology, 5 (1967), 32-37.

II V. M. Satir, Conjoint Family TheraPy: AGuide to Theory and Technique (Palo Alto,Calif.: Science and Behavior 'books, 1964); LNavran, "Communication and Adjustment InMarriage," Family Process, 6 (1967). 173.184:and A. I. Ferreira and W. D. Winter, l'Information Exchanm and Silence In Normal andAbnormal Pam es," Family Process, 7 (1968),251-276.

bination of game behavior and visualnonverbal cues induces more cooperativebehavior than does game behavioralone," the combination of game behav-ior, words, and Voice tone and inflectioninduces more cooperative behavior thaneither game behavior or game behaviorand visual nonverbal cues," and thecombination of game behavior, words,and all nonverbal cues induces more co-operative behavior than either gamebehavior, game behavior and notes,game behavior and visual nonverbalcues, or game behavior, words and voicetone inflections." These studies are con-gruent with the findings of studies of per-suasion, which have indicated that face-to-face communication is most effectivein producing attitude and behavioralchanges." Thus the greater number ofchannels through which messages canbe sent, the simultaneous redundancyprovided by multi-channel communica-tion, and the two-way interaction pro-vided in face-to-face discussion, all in-crease the effectiveness and clarity ofmessages aimed at inducing cooperativebehavior. It may be that the morechannels used to send messages the lessthe ambiguity concerning the meaningand motivating intentions of the mes-sages (given simultaneous redundancy).Visual nonverbal cues, for example, may

52 H. Wlchman, "Effects of Isolation andCommunication on Cooperation in a Two-Person Game," Journal of Personality and SocialPsychology, 16 (1970), 114-120; and ._%V. Dorris,G. Gentry, and H. H. Kelley, "Effects UponBargaining of Modality of Interaction, InitialDegree of Conflict, and Initial Orientations,"Journal of Personality and Social Psychology,in press.

53 Wlchman.54D. S. Ellis, "An Analysis of the Differential

Effects of Various Types and Degrees of Corn-muntcation Opportunity on Conflict BetweenGroups," Disc Purdue Univ. 1965; D. H. Smith,"Communication and Negotiation Outcome,"Journal of Communication, 19 (1969), 248-256;and Wichman.

55E. Katz and P. F. 1Lararsfeld, Personal In-fluence: The Part Played by People in the Flowof Mass Communications (Glencoe, 1114 FreePress, 1955).

76 SPEECH MONOGRAPHS

be the most ambiguous message and,while they do increase the probability ofcooperative behavior, they are not aseffective as more unambiguous words.The least ambiguous situation is whenall the nonverbal cues, words, and gamebehavior indicate cooperative intent andinformation about cooperative behav-ior, and it is this situation which doesinduce the most cooperative behavior.

Message Content

Effectiveness in inducing cooperativebehavior increases as the messages con-tain explanations of how to behave co-operatively and a rationale for such be-havior." Appeals emphasizing the needfor cooperation, the other's responsibilityto engage in cooperative behavior, anda , request for help in creating a co-operative situation are productive ofcooperative behavior 51 Indicating one'scooperative intentions by proposing com-promises," accurately paraphrasing theother's position and feelings," inducingthe other to role reverse by paraphrasingone's position and feelings," and non-verbally expressing coldness or an abruptswitch from anger to warmth" all in-

85 Ellis.51 J, W. Dorris, "Reactions to Unconditional

Cooperation: A Field Study Emphasizing Vari-ables Neglected in Laboratory Research, Jour-nal of Personality and Social Psychology, 22(1972), 387.397.

58 Krauss and Deutsch, "Communication inInterpersonal Bargaining"; and D. W. Johnson,"Effects of Warmth of Interaction, Accuracy ofUnderstanding, and the Proposal of Comprom-ises on the Listener's Behavior," Journal ofCounseling Psychology, 18 (1971), 207-216.

50 D. W. Johnson, "Use of Role Reversal inIntergrbup Competition," Journal of Personalityand Social Psychology, 7 (1967), 135.141; and"Warmth of Interaction, Accuracy of Under-standing, and Compromises."

so Johnson, "Use of Role Reversal in Inter-goup Competition," (1067); and "Effectivenessof, Role Reversal: Actor or Listener," Psycho.logical Reports, 28 (1971), 215-282.

si Johnson, "Warmth of Interaction, Accuracyof Understanding, and Compromises"; and"Effects of the Order of Expressing Varrnth andAnger on the Actor and the Listener." Journalof Counseling Psychology, 18 (1971), 571-578.

crease the probability of successfully in-ducing cooperative behavior. It is onlywhen these factors are tad competently,however, that they will be effective."The similarity of connotative meaningsof concepts and words relevant to theconflict is important in successful con-flict management."

Situational Varidbles

The ability to survey another's post-cooperative-agreement behavior is im-portant if one is to honor his commit-ment to behave cooperatively." It isevident that communication must takeplace under a promotive goal interde-pendent structuie in which cooperationis necessary to achieve rewarding out-comes if communication is to facilitatecooperation." The affective relationshipof liking or friendship makes a differencein how bargaining is conducted and thetype of agreements made." Psychopath-ology of the individuals involved in theconflict seems to interfere with usingcommunication to agree upon coopera-tive behavior, as individuals who havepathological tendencies tend to mis-trust and violate such agreements."

CONCLUSIONS

The review of research on communi-cation in conflict situations evokes sev-eral criticisms which can be made con-

62 D. W. Johnson, "The Use of Role-Reversalin Intergroup Competition," Diss. ColumbiaUniv. 1966; and "Use of Role Reversal" (1067).

83 M. Katz, "Agreement on Connotative Mean-ing in Marriage,' Family Process, 4 (1965), 64-74; and H. C. Triandis, "Cognitive Similarityand Communication in a Dyad," Human Rela-tions, 13 (1960), 115-183.

64 Rixenstine et al., "Collaboration."65 M. Sherif, In Common Predicament: Social

Psychology of Intergroup Conflict and Cooper-ation, International Series in the BehavioralSdences, ed. J. E. Horrocks (Boston: Houghtonhlifflin, 1966)._

R. Morgan and J. Sawyer, "BaNaining,Expectations, and the Preference for EqualityOver Equity," Journal of Personality and SocialPsychology, 6 (1967), 139-149.

67 Bixenstine and Douglas.

COOPERATIVE BEMIS/10R CONFLICTS -77

cerning almost all of the research. Thefirst criticism focuses on the lack of con-ceptualization of conflict and communi-cation. Often the most superficial con-ceptual definition is given for conflictand often no conceptual definition at allis given for comunication. There havebeen instances where a standard opera-tional definition has provided an areawith some momentum, as with the IQtest and the F-Scale, but in generalproductive research depends upon ade-quate conceptualization of the conceptsbeing investigated. Without adequateconceptualization it is difficult to buildtheoretical models to organize and stim-ulate research. Concepts determine thebehavioral field observed, affecting theprinciples derived from the observations,which in turn affect the hypotheses,laws, and theories constructed. Withoutclarification of the concept "communica-tion," no theory of effective communica-tion in conflict situations can be built.

The second general criticism is thelack of theorizing to test significant hy-potheses. Although many of the studiesfind reliable relationships among vari-ables they do not demonstrate the con-ditions under which such relationshipsAre strongest or the reasons why suchrelationships exist. A variety of post hocexpipnations are presented to explain theresults of individual studies, but a fan-tastic lack of actual theory is beingtested. The value of research is to pro-mote theoretical advances in the areaunder study. The relationship betweentheory and research is complementary;theory is not useful unless it is empiri-cally verified by research, and researchis not useful unless it is related to theory.It is their relationship to theory whichgives research findings their significance;research findings which have no relationto theory are trivial. From this point ofview almost all research conducted oncommunication in conflict situations is

trivial and of no significance as yet. Al-though there are numerous studies inthe area, the lack of theory leaves theresearch findings unorganized, undirec-ted, unexplainable, and unimportant.

Two major criticisms can be madeabout how communication has beenoperationalized in conflict research. First,the operational linitions used for com-munication have been inadequate in

clarifying the concept of communication.When studying an inexact, complex con-cept such as communication the opera-tional d.finitions should explicate com-munication into a concept which canbe more precisely specified in order tomake future, research more theoreticallysignificant. This has not been done inthe conflict research. Second, the opera -tionalizations of communication havebeen unproductive in advancing theory.The purpose of an operational definitionis to serve as an index of a conceptwhich is related to other concepts withina theory. Operational definitions servethe purpose of defining or partially de-fining theoretical concepts in terms ofobservable data in order to test the em-pirical validity of a theory. The lack ofconceptualization of communication andthe inadequacy of the operational defi-nitions used combine to make most ofthe research in this area unproductive inadvancing theory.

A fourth general criticism is that thedependent variable most often used isthe quantity of cooperative behavior inthe conflict situation. It is doubtfulwhether this is the most appropriatedependent variable to use. There is a

large gap between accurately receivinga cooperative message and deciding torespond with cooperative behavior.

Research on communication in conflictsituations has by and large ignored thetheoretical literature in communication.This results in misdirected researchwhich is irrelevant to current communi-

78 SPEECH MONOOR4PHS°Y.

cation theory, repeats work that has al-ready been done, and results in dubiousassumptions about the communicationprocess, all of which may invalidatemuch of the research which has beenconducted.

Finally, there is freqnently an un-bridled gap between the results foundby researchers and the application ofthe findings to other conflict situations.The social importance of this area makesit imperative that resea'rch be conductedto validate communication procedureswhich can be used in actual interpersonaland intergroup conflicts,

The research findings reviewed 'can beused as evidence that a variety of mes-sage and channel characteristics; situa-tional variables, and communication pro-cedures are effective in inducing coopera-tive behavior in a conflict situation. Yetmost of the research reviewed is only oflimited value for validating theory asonly post hoc explanations of the resultscan be presented. Most of the research

reviewed has not contributed significant-ly to an understanding of the conditionsunder which certain types of messagessent through certain types of channelsin the context of certain situational vari-ables will be received in such a way asto influence the decision to respond co-operatively. Future research in this im-portant area should be characterized bythe testing of theoretically based hypo-theses drawn from models of the com-munication process which include con-cepts that readily lend themselves toclarification and explication throughoperationalization. Most of the researchreviewed here has used communicationvariables as independent variables; inthe future it may be fruitful to investi-gate the conditions under which coopera-tively oriented messages will be sent toan opponent. Finally, more time and ef-fort should be spent on validating pro-cedures of effectively communicatingwithin a variety of interpersonal and in-tergroup conflicts.

SPECIAL REPORTSOPPORTUNITY TO COMMUNICATE AND SOCIAL

ORIENTATION IN IMAGINARY-REWARD BARGAINING

JAMES' GREENWOOD

GENERALLY, the assumption ismade in the traditional approach

to argumentation and bargaining thatcommunication is always of value in con-flict resolution. The task has been de-fined as prescribing the most effectivimeans of communicating. This study ap-proachei conflict resolution in a descrip-tive rather than a prescriptive way andtests the assumption that communicationhas inherent value in conflict resolution.

Surprisingly negative results have oc-curred in some bargaining studies com-paring the presence and absence of op-portunity to communicate. Deutsch andKrauss, Wandell, and Scodel et al., re-ported little or no improvement in theperformance of subjects who were per-mitted explicit communication comparedwith subjects who were not allowed tocommunicate.' Communication actuallyled to lower levels of cooperation in thefirst two studies. In studies by DLnielsand by Swensson, subjects with the op-portunity to communicate tended to per-form more successfully than subjectswithout that opportunity, although dif-ferences did not reach the .05 level ofsignificance.5

Mr. Greenwood is a doctoral student in the De.partment of Communication, the Ohio StateUniversity.

Morton Deutsch and Robert M. Krauss,"Studies of Interpersonal Bargaining," Journalof Conflict Resolution, 6 (196'4, 52.76; WilliamAllen Wandell, "Group Membership and Com.munication in a Prisoner's Dilemma Setting."Diu. University of Houston 1967; Alvin Scodel,J. Sayer Minas, Philburn Ratoosh, and MiltonLipett, "Some Descriptive Aspects of TwoPersonNon-Zero-Sum Games," Journal of ConflictResolution, 3 (1959), 114.119.

2 Victor Daniels, "Communication, Incentive,

SPEECH MONOGRAPHS, Vol. 41, March 1974

However, those studies focusing on thecontent of messages rather than on themere opportunity to communicate pre-sent a different picture. Krauss andDeutsch, and Loomis reported statis-tically significant improvement whensubjects were tutored to engage in fairbargaining or when .sObjetts- increasedthe completeness of their messages byincluding statements of expectation, in-tention, retaliation, and absolution .5But Swensson failed to find a differencein the effects of three messages with dif-fering affective tones.4

Varying social orientation, Deutschfound subjects with a cooperative orien-tation behaved cooperatively in a pris-oner's dilemma significantly more fre-quently than did subjects with an in-dividualistic or competitive orientation.Further, subjects who were allowed tocommunicate tended to cooperate morefrequently than did those without theopportunity to communicate, althoughsignificant differences attributable to op-portunity to communicate occurred onlyfor subjects with an individualistic orien-tation.5

and Structural Variables in interpersonal Ex-change and Negotiation," Journal of Expertmental Social Psychology, 3 (1967), 47.74; Richand G. Swensson. "Cooperation in the Prisoner'sDilemma Game I: The Effects of AsymmetricPayoff Information and Eiplicit Communica-tion," Behavioral Science, 12 (1967), 314.322.

3 Robert Krauss and Morton Deutsch, "Com-munication in Interpersonal Bargaining," Jour-nal of Personality and Social Psychology, 4(1966), 572.577; James L. Loomis, "Communi-cation, the Development of Trust, and Coopera-tive Behavior," Human Relations, 12 (1959),305-315.

4 Swensson, 314-322.Morton Deutsch, "The Effect of Motiva-

80 SPEECH .MONOORAPHS

The two variables manipulated in thisstudy were social orientation and theopportunity to communicate. Two socialorientations were used: (1) cooperativeeach subject was led to feel that thewelfare of the other person as well as hisown welfare was important and that theother person felt the same way; and (2)competitiveeach subject was led to feelthat he wanted to do as well as he couldfor himself and also better than the otherperson and that the other person feltthe same way .° Three levels of communi-cation opportunity were employed: (1)unrestrictedboth subjects could talkto the_ other as often as they- wished;(2) moderate restrictionboth subjectscould talk to the other three times ineach game; and (3) high restrictionboth subjects could talk to the otheronly once in each game. Only the fre-quency of opportunity to communicatewas controlled, not length of messagenor type of message, The dependentvariable was the number of "success-ful" bargains each pair of subjects con-cluded.

Two hypotheses were tested: (1) dif-ferences in the opportunity to communi-cate will affect the number of successesobtained in a simple bargaining setting;and (2) cooperative orientation andcompetitive orientation will affect thenumber of successes obtained,

METHOD

The game matrix developed to providethe essential features of bargaining isshown in Figure 1.7 Player' row can

Hotta! Orientation Upon Trust and Suspicion,"Human Relations, 13 (1960), 123439

()The Instructions to the subtects includeda slight modification of the instructions Deutschused to establish these orientations. Sec Deutsch,123-139.

7 A game matrix is an application of thetheory of games of strategy which is enjoyingIncreasing popularity with political scientists,economists, and the military. See. e.g., ThomasC. Schetling, The Strategy of Conflict (Cam-

choose either row A or B and playercolumn can choose column X, Y, or Z,

XColumn

Row A $20,13B $1343,

$147 $033$3,$13 $SP

'First figure In each cell is row's payoff.

Ficultx 1. Payoff Matrix

The cell intersecting row's choke andcolumn's choice is the payoff, with rowreceiving the first number in the pair.A "successful" resolution of the conflictis defined as cell AX, the cell with thegreatest total payoff This is theinst...solution for both players since it paysa total of $23, or $7 more than the nextbest cell, If it is to be the best solutionfor column, however, partners mustagree to redistribute, the payoff; as origi-nally distributed, 'column could nor rea-sonably be expected to choose X, sincehis payoff would inevitably be higher inY. Thus, without an agreement to re-distribute, an AX combination must beconsidered an error attributable to afailure to understand the game.

With this matrix, both players shouldperceive the possibility of reaching anagreement in which each party would bebetter off than if no agreement is reached.Without an agreement to split the $23in cell AX, the rational choice for columnis Y. The rational choice for row, whorealizes that column will choose Y, is Aif row wishes to minimize the differencebetween himself and column. Otherwise,row's rational choice is B. Hence, eithercell AY or cell BY can be consideredrational in the absence of a redistributionagreement, depending on row's. goals.

bridge: Harvard Univ, Press, 1960); MartinShubik, ed., Game Theory and Related Ap.proaches to Social Behavior (New York:1964): Anatol Rapoport, Fights, Games, andDebates (Ann Arbor: Univ. of Michigan Press,1960).

SPECML

Neither player has a disproportionateamount of power. Column is guaranteeda minimum of $7 in Y, so he is not forcedto accept an "unfair" bargain in cell AX,Row controls $20 in cell AX. Therefore,he has sufficient wherewithal to make anenticing offer (at least better than $7)to column. Thus, both players wouldbenefit from an arrangement distributingthe lucre in cell AX.

Subjects, 84 volunteers from speechfundamentals classes at Northern IllinoisUniversity, were assigned to pairs, eachpair playing a series of seven trials.Seven pairs of subjects were assignedto each of the six social orientation/com-munication restriction cells. Cooperative-ly oriented subjects always played withtheir like orientation, as did competi-tively oriented subjects.

Subjects were instructed that the pay-off within any cell was subject to rear-rangement if both players agreed to thenew distribution. This provided thesource of conflict and the subject matterfor discussion between players, Any pairunable to reach a mutually acceptableredistribution submitted secret ballots,

REPORTS 81

thus accepting as a payoff the intersec-tion of row and column without a redis-tribution, All payoffs were imaginary,and subjects were so informed beforeplaying.

REavvrs

As Table 1 shows, main effects werefound for both social orientation andcommunication opportunity. The inter-action was not significant. The patternof means (Table 2) indicates that suc-cessful bargaining is facilitated by acooperative orientation and by increasedopportunity_ to communicate. Jlowevert-tests among pairs of cell means withinorientation and communication condi-tions revealed only two differences sig-nificant beyond .05: among subjects witha competitive orientation, unrestrictedcommunication results in significantlymore successful bargains than does high-ly restricted communication; and, givenunrestricted communication, a coopera-tive social orientation results in signifi-cantly more successful bargains than doesa competitive orientation.

TABLE 1ANALYSIS Or VARIANCE SUMMARY TABLE

Source SS di MS

Rows (orientation) 50.38 50.38 15.13Columns (communication) 26.33 2 12.16 3.95Interaction 1.48 2 .74 .22Error (within cells) 119.72 36 3.33

Totals 197.91 41

oSignificant beyond .05 level,

TABLE 2MEAN SUCCESSES PER TREATMENT"

Opportunity to CommunicateModerate High

Unrestricted Restriction Restriction Combined

OrientationCompetitive 2.71t. 2.00 1.14 1,95Cooperative 5.42f 3.86 3.14 4.14Combined 4.06 2.93 2.14

'Means with same sign differ beyond .05 level.tMeans with same sign differ beyond .05 level.

"Means were compared statistically only within orientation and communication levels.

82 SPEECH MONOGRAPHS

AN EXPERIMENTAL VERIFICATION OF 4CHELLINGSTACIT COMMUNICATION HYPOTHESIS

THOMAS E. HARRIS and ROBERT M. SMITH

CHELLING isolates the concept"tacit communication," communica-

tion occurring via the common under-standings of the two parties rather thanby means of explicit messages sentthrough identifiable channels.' He as-serts that such communication must oc-cur to coordinate behavior whepeverother messages are incomplete, difficult,unreliable, or nonexistent. Although onemay argue that it is fairly obvious that"a great deal of communication goes onwithout any spoken or written state-ments being made,"2 the nature andscope of this form of communicationhave not been carefully studied.3

Tacit communication depends on thesaliency of certain cultural or situationalfeatures in a conflict or, for that matter,in any setting where explicit messagesare not feasible. Bargains frequently canbe struck without overt communication,simply by the tacit observation by bothparties of some situational element thatstands out in such a way that its mutualobservation becomes likely. For example,two parties faced with the need to divide$100 without overt communication' willalmost always utilize a SO-50 split. Ahusband and wife accidentally separated

Mr. Harris is assistant professor of speech,Rutgers University. Mr. Smith is assistant pro.fesor of speech, Wichita State University. Theywish to acknowledge the assistance of HerbertW. Simons.

Thomas C. Schelling, The Strategy of Cowflict (Cambridge: Harvard Univ. Press, 1960).

2 Michael Nicholson, Conflict Analysis (NewYork: Barnes and Noble, 1971), p. 3.

8Iames T. Tedeschl, "Threats and Prom-ises," p. 160, and Bertram H. Raven and ArieW. Kruglanski, "Conflict and Power," p, 91, inThe Structure of Conflict, ed. Paul Swingle(New York: Academic Press, 1970).

in a department store normally manageto accomplish reunion without intercomsor paging systems simply by returningto their most frequented section of thestore or going to the information booth.As Schelling says:

The study of tacit bargainingbargaining inwhich conttnunitaiiok As intomPlce 0* ,407possibleassumes importance .. in connectionwith limited war, or, for that matter, withlimited competition, Jurisdictional maneuvers,Jockeying in a traffic Jam, or getting along witha neighbor that one doeS not speak W.4

In spite of the concept's importance,

except for anecdotes and' homely illustrations,we suddenly become aware that we haie verylittle evidence: finished research that focusessystematically on historical cases of such phe-nomena as tacit bargaining.8

Three studies, besides Schelling's in-formal ones, have been conducted -em-ploying Tacit Communication Games(TCG). Willis and Joseph used TCGin continual play among various partnercombinations and concluded that con-tinual play tended to increase informa-tion but decrease coordination.° Solomonfound that schizophrenics were less suc-cessful in TCG than were college stu-dents.? Fry found that success in TCG

4 Schelling, p. 53.5 Charles A. McClelland, "The Reorientation

of the Sociology of Conflict: A Review," rev. ofThe Strategy of Conflict, by Thomas C. Schell.ing, and Conflict and Defense: A GeneralTheory, by Kenneth E. 13oulding, Journal ofConflict Resolution, 6 (1962), 92.

Richard H. Willis and Myron L. Joseph,"Bargaining Behavior. I. 'Prominence' as a Pre-dictor of the Outcome of Games of Agreement,"Journal of Conflict Resolution, 3 (1959), 102.113.

I L. Solomon, "Schizophrenic Communica

SPECIAL REPORTS

is partly a function of age, with thegreatest success achieved by college stu-dents, followed by adolescents, followedby pre-adolescents.8

METHOD

Subjects were 74 undergraduates at alarge university and 66 undergraduatesat a small college. At both schools, sub-jects were randomly divided into twogroups. MemberS of the "real partners"group were randomly assigned to dyads;members of the "hypothetical partners"group were told that they were playingwith an imaginary partner similar tothemselves. Instructions for all subjectswere:

During the next fifteen to twenty minutes, wewill be conducting an exercise in the formof a game designed to investigate one aspect ofcommunication. This is not a test and eachperson's answers will be kept strictly confi-dential. Please refrain from talking during theexerdse. As the instructions I am about to handyou make clear, this is a game of coordinationand you can win only if you and your partnerare able to coordinate your answers. Please readthe instructions carefully. After completing thefirst page, you may ask questions concerning theprocedure.

Each copy of the TCG contained gen-eral instructions explaining the conceptof coordination as an exercise in non-verbal communication. The "real part-ners" group was also asked to indicateon a seven-level scale the amount ofprior interaction they had had with theirrandomly assigned partners. As a sampletest to familiarize subjects with theTCG, the following example was used:"Name 'heads' or 'tails.' If you and yourpattner name the same side, you bothwin a prize." The vast majority picked

Hon: Experimental Studies in Tacit Coordina-tion," paper read at Eastern Psychological As-sociation, New York, April 1960.

8 Charles L. Fry, "A Developmental Examin.ation of Performance in a Tacit CoordinationGame Situation," Journal of 13,rsonolay andSocial Psychology, 5 (1967), 227-281.

83

heads and, apparently understood theimportance of coordinated behavior forwinning the gime.

Ficuaz 1.Tacit Communication Game

I. Circle one of the numbers listed below. Youwin if you and your partner succeed in circlingthe same number.

7 100 13 261 99 555

2. Put a check mark in one of the 12 boxes(). You win if you and your partner bothsucceed in checking the same box.

n O n

11 O n

O n3. Circle one of the amounts of money listedbelow. You win if you and your partner suc-ceed in circling the same amount of money.

$10 $64,000 $10,000 $1,000,000 $64 $8000

4. You are to meet your partner at GrandCentral Station in New York City, but you andyour partner do not know the hour of themeeting. You have no prior understandingwith your partner on when to meet and youcannot communicate with each other. You bothmust guess the exact minute of the day forthe meeting. At which of the times listed be-low will you appear for the meeting? Circleone.

9:30 a.m. 3:00 p.m. 12:00 noon 5:00 p.m. 10,00 a.m.

1:00 p.m. 9:00 a.m. 2:00 p.m.

5. You and your partner are each to pick oneof five letters: K, G, W, L, or R. If you pickthe same letter, you win; if you pick differentletters, you get nothing. The prize you getdepends on the letter you both pick; but theprizes are not the same for each of you, andthe letter that would yield you the highestprize may or may not be his most profitableletter. For you the prizes would be as follows:K =4; G = 3; W = I; L =2; R = 5.

You have no idea what his schedule of prizeslooks like. You begin by proposing to him theletter Rthat being your best letter. Beforehe can reply, the master of ceremonies inter-venes to say that you were not supposed to beallowed to communicate and that any furthercommunication will disqualify you both. Youmust simply circle one of the letters below,hoping that the other chooses the same letter.Which letter do you pick?

6. You and your partner are each given a pieceof paper, one of which is blank and the otherwith an "X" written on it. The one who gets

84 SPEECH MONOGRAPHS

the "X" has the choice of leaving it alone orerasing it; the one who gets the blank sheethas the choice of leaving it blank or writing an"X" on ft. If when you have made your choke,there is an "X" on only one of the sheets, theholder of the "X" gets $3.00 and the holder ofthe blank sheet gets $2.00. If both sheets have"X's" or both sheets are blank, neither getsanything. Your sheet of paper has the original"X" on it do you leave it alone or erase it?

Check one:leave "X" alone erase the "X"

Six of the TCG exercises developed bySchelling were used. As Figure 1 indi-cates, the first four were entirely testsof cooperative behavior while the lasttwo introduce a mixed-motive or bar-gaining situation where one player canwin at the relative expense of the other.

RESULTS

Schelling's predictions proved to besubstantially correct. The answers heforecast were: (1) the first three num-bers with 7, 100 and 13 in order of popu-larity; (2) upper left-hand corner; (3)$1,000,000 or some number to the pcwerof ten (I tested against the first pre-diction); (4) twelve noon; (5) R; and(6) "X."

The "real partners" and "hypotheticalpartners" groups did not differ, andtheir results are reported together. Fur-

thermore, in the "real partners" group nosignificant correlation was found betweenextent of coordination and reported pre-vious interaction with partner.

Table 1 shows that extensive tacit com-munication occurred, with nine of twelvecomparisons confirming Schelling's pre-dictions via chi square tests. Two of the

TABLE ICut SQUARE RESULTS FOR CONFORMITY

TO SCHELLING'S PREDICTIONS

Question College

1 .004112 .0000l3 .129084 .0046705 .129846 .21198

Significant beyond .05 level.

University

X95

.00424

.000930

.014416

.01330

three comparisons that failed to verifythe predictions involved students at thesmall college in the two mixed-motiveexercises.

The results underscore the importanceof tacit communication in situations in-volving conflict. By understanding tacitvariables, opponents are able to gain in-formation about the other's present andfuture behavior. Such knowledge worksboth to enhance strategy and to avoidtotally dysfunctional escalation.