a sequential functional model of nonverbal...

19
Psychological Review 1982, Vol. 89, No. 3, 231-249 Copyright 1982 by the American Psychological Association, Inc. 0033-295X/82/8903-0231J00.75 A Sequential Functional Model of Nonverbal Exchange Miles L. Patterson University of Missouri—St. Louis A sequential functional approach to understanding and analyzing nonverbal ex- change is proposed. The sequential analysis of nonverbal exchange begins with the identification of a set of antecedent factors (personal, experiential, and re- lational-situational) that initiate preinteraction mediators. These mediators, in the form of behavioral predispositions, potential arousal change, and cognitive- affective assessment, structure both the perceived functions of an interaction and the levels of nonverbal involvement initiated by each interactant. In contrast to other theories of nonverbal exchange, which explicity or implicitly focus on the intimacy function in interaction, this model proposes that the functional bases of interaction are variable. Of particular importance are the distinctions proposed among the intimacy, social-control, and service-task functions. The activation of different functions in an interaction is assumed to result in differing arousal, cognitive, and behavioral patterns in the interactants. Nonverbal exchange is predicted to be more unstable as the interactants' perceived functions and ex- pectancies are more divergent. The model also proposes that, especially for the social-control function, one's nonverbal behavior may be a product of a general strategy or script and not simply a reactive product of the other person's preceding behavior. Although the amount of empirical re- search on nonverbal behavior in social in- teraction has been voluminous, theoretical formulations have been relatively sparse. Perhaps the most influential theoretical po- sition in this area of research has been Ar- gyle and Dean's (1965) equilibrium model of intimacy. The basic assumption of equi- librium theory is that there is a definable level of intimacy appropriate to any given interaction. That intimacy is manifested by behaviors such as interpersonal distance, gaze, and smiling and by the intimacy of the topic of conversation. If the intimacy equi- librium is disturbed by a change in any one or more of the other component behaviors, An earlier, briefer version of this paper was presented at the annual meeting of the American Psychological Association, Montreal, September 1980. Preparation of this paper was supported by a National Institutes of Health Grant USPH 5 R01 MH 32386-02 awarded to the author. I am grateful to Dianne Patterson and Bill Ickes for their comments on an earlier draft of this pa- per. Comments from two anonymous reviewers were especially helpful in developing the final version of this paper. Requests for reprints should be sent to Miles L. Pat- terson, Department of Psychology, University of Mis- souri, St. Louis, Missouri 63121. then a compensatory adjustment in one or more of the other component behaviors would serve to restore the intimacy equilib- rium. For example, a too close approach ini- tiated by one member of a pair might pro- duce a decrease in gaze and smiling by the other member of the pair. General support for the compensation process predicted by equilibrium theory has been found across a variety of different studies (see reviews by Cappella, 1981; Patterson, 1973). However, some results showing the di- rectly opposing pattern of adjustment, rec- iprocity, or matching intimacy input (e.g., Breed, 1972; Chapman, 1975; Jourard & Friedman, 1970) pose a problem for equi- librium theory. In attempting to reconcile the divergent findings in a more comprehen- sive framework, I developed an arousal model of intimacy exchange (Patterson, 1976). This theory proposes that behavioral adjustments to changes in intimacy are me- diated by arousal change and cognitive la- beling processes. Thus, it was predicted that sufficient changes in one person's intimacy behavior would precipitate arousal change in the other person. Depending on the re- lationship, situation, and other cognitive 231

Upload: dinhnhi

Post on 10-Jun-2018

233 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

Psychological Review1982, Vol. 89, No. 3, 231-249

Copyright 1982 by the American Psychological Association, Inc.0033-295X/82/8903-0231J00.75

A Sequential Functional Model of Nonverbal Exchange

Miles L. PattersonUniversity of Missouri—St. Louis

A sequential functional approach to understanding and analyzing nonverbal ex-change is proposed. The sequential analysis of nonverbal exchange begins withthe identification of a set of antecedent factors (personal, experiential, and re-lational-situational) that initiate preinteraction mediators. These mediators, inthe form of behavioral predispositions, potential arousal change, and cognitive-affective assessment, structure both the perceived functions of an interaction andthe levels of nonverbal involvement initiated by each interactant. In contrast toother theories of nonverbal exchange, which explicity or implicitly focus on theintimacy function in interaction, this model proposes that the functional basesof interaction are variable. Of particular importance are the distinctions proposedamong the intimacy, social-control, and service-task functions. The activationof different functions in an interaction is assumed to result in differing arousal,cognitive, and behavioral patterns in the interactants. Nonverbal exchange ispredicted to be more unstable as the interactants' perceived functions and ex-pectancies are more divergent. The model also proposes that, especially for thesocial-control function, one's nonverbal behavior may be a product of a generalstrategy or script and not simply a reactive product of the other person's precedingbehavior.

Although the amount of empirical re-search on nonverbal behavior in social in-teraction has been voluminous, theoreticalformulations have been relatively sparse.Perhaps the most influential theoretical po-sition in this area of research has been Ar-gyle and Dean's (1965) equilibrium modelof intimacy. The basic assumption of equi-librium theory is that there is a definablelevel of intimacy appropriate to any giveninteraction. That intimacy is manifested bybehaviors such as interpersonal distance,gaze, and smiling and by the intimacy of thetopic of conversation. If the intimacy equi-librium is disturbed by a change in any oneor more of the other component behaviors,

An earlier, briefer version of this paper was presentedat the annual meeting of the American PsychologicalAssociation, Montreal, September 1980. Preparation ofthis paper was supported by a National Institutes ofHealth Grant USPH 5 R01 MH 32386-02 awarded tothe author. I am grateful to Dianne Patterson and BillIckes for their comments on an earlier draft of this pa-per. Comments from two anonymous reviewers wereespecially helpful in developing the final version of thispaper.

Requests for reprints should be sent to Miles L. Pat-terson, Department of Psychology, University of Mis-souri, St. Louis, Missouri 63121.

then a compensatory adjustment in one ormore of the other component behaviorswould serve to restore the intimacy equilib-rium. For example, a too close approach ini-tiated by one member of a pair might pro-duce a decrease in gaze and smiling by theother member of the pair. General supportfor the compensation process predicted byequilibrium theory has been found across avariety of different studies (see reviews byCappella, 1981; Patterson, 1973).

However, some results showing the di-rectly opposing pattern of adjustment, rec-iprocity, or matching intimacy input (e.g.,Breed, 1972; Chapman, 1975; Jourard &Friedman, 1970) pose a problem for equi-librium theory. In attempting to reconcilethe divergent findings in a more comprehen-sive framework, I developed an arousalmodel of intimacy exchange (Patterson,1976). This theory proposes that behavioraladjustments to changes in intimacy are me-diated by arousal change and cognitive la-beling processes. Thus, it was predicted thatsufficient changes in one person's intimacybehavior would precipitate arousal changein the other person. Depending on the re-lationship, situation, and other cognitive

231

Page 2: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

232 MILES L. PATTERSON

cues, that arousal change will be labeled asa positive or negative feeling state. Thisarousal model predicts that negatively la-beled feeling states (e.g., fear, anxiety, em-barrassment) facilitate the compensatoryadjustments described in equilibrium theory.In contrast, positively labeled feeling states(e.g., liking, love, relief) facilitate the recip-rocation or enhancement of intimacy. Fi-nally, when the initial intimacy change is notsufficient to precipitate arousal change, nobehavioral adjustments will be predicted.Thus, the arousal model provides an explan-atory mechanism covering the full range ofadjustments in nonverbal intimacy.

The results of a few studies over the lastseveral years have provided support for thepredictions of the arousal model (Foot,Chapman, & Smith, 1977; Foot, Smith, &Chapman, 1977; Whitcher & Fisher, 1979).In general, those results showed that recip-rocation of intimacy was related to positivelyvalenced arousal change and compensationfor intimacy was related to negatively va-lenced arousal change. Presumably, the self-labeling of the arousal change (Schachter& Singer, 1962) was the mechanism deter-mining the positive or negative feeling state.However, Ellsworth (Note 1) noted thatsuch a cognitive self-focus may be a luxuryin the typical interaction in which the indi-vidual must be concerned about the motivesof the other person. That is, the cognitivework, whether it is precipitated by. arousalchange or not, will commonly involve attri-butions about others more frequently thanit will about oneself. In addition, some of ourown recent research has not clearly sup-ported the predictions of the arousal model(Patterson, Jordan, Hogan, & Frerker,1981). Those results will be discussed insome detail later. However, at this point itis probably sufficient to note that a predicteduniform pattern of increased arousal to dis-crete touch was not found in that study.

If the cognitive and arousal mediators ofnonverbal exchange are, in fact, more vari-able than the arousal model suggests, a sub-stantially different framework may be neededto accommodate greater diversity in themediation and outcome phases of nonverbalexchange. Ellsworth's (Note 1) emphasis onother-directed attributions and Knowles's

(Note 2) suggestion that cognitions may beprimary and arousal secondary in mediatingnonverbal exchange provide direction forsuch a theoretical framework. Specifically,both of these proposals seem to converge onthe importance of the meaning of the otherperson's change in nonverbal intimacy. Oneuseful way of classifying that meaning is interms of the functions served by nonverbalbehavior in interactions. The implication inapplying a functional approach to explainingnonverbal exchange is that different func-tional bases will lead to different patterns ofexchange. Fortunately, functional classifi-cations have already been advanced in var-ious analyses of nonverbal behavior (e.g.,Argyle, 1972; Argyle & Dean, 1965; Argyle,Lalljee, & Cook, 1968; Ekman & Friesen,1969; Harrison, 1973; Kendon, 1967). Com-mon to most of these classifications are thefollowing functions: (a) communication, (b)regulating interaction, and (c) expressingintimacy.

The purpose of this paper is to develop amore detailed functional classification as thebasis for a new and more comprehensivemodel of nonverbal exchange. Ideally, thisnew model should be able both to explainthe phenomena covered by earlier modelsand to account for additional results incon-sistent with the predictions of the earliermodels. In laying the groundwork for thisnew theory, it will be necessary first to dis-cuss the focal domain of behaviors and thento specify the functional classification basicto its operation.

Nonverbal Involvement

The first issue may be described as a def-initional one involving a distinction betweenthe behaviors of interest and the functionsserved by those behaviors. This might seemlike a trivial concern, but the assumptionsresulting from the use of particular terms orlabels can substantially affect our approachto researching various issues. This is espe-cially true in the case of the extensive re-search on nonverbal intimacy. The term in-timacy has been used to describe both a setof behaviors (distance, gaze, smiling, andverbal intimacy) and a function served bythose behaviors (Argyle & Dean, 1965; Pat-

Page 3: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 233

terson, 1976). Such a dual usage is commoneven though it is readily acknowledged thatthe referent behaviors can also serve func-tions other than those of expressing inti-macy. However, the likely result of that ter-minology is decreased sensitivity to alternatefunctions served by those behaviors.

One solution will be the adoption of amore neutral^et appropriate, label for thebehaviors of interest. Specifically, I wouldpropose the term nonverbal involvement be-haviors to describe a set of behaviors oper-ationally defining (in some as yet unknown,weighted fashion) the degree of involvementmanifested between individuals in a socialsetting. As a behaviorally linked construct,nonverbal involvement is conceptually dis-tinct from the higher order, functional con-structs such as intimacy. A tentative listingof behaviors appropriate for involvement willinclude the following: (a) interpersonal dis-tance, (b) gaze, (c) touch, (d) body orien-tation, (e) lean, (f) facial expressiveness, (g)talking duration, (h) interruptions, (i) pos-tural openness, (j) relational gestures,1 (k)head nods, and (1) paralinguistic cues suchas intonation, speech rate, volume, and soforth.2 The first five behaviors coincide withMehrabian's (1969) list of immediacy cues.Mehrabian defined immediacy as "the ex-tent to which communication behaviors en-hance closeness to and nonverbal interven-tion with another" (1969, p. 203). Conceptu-ally, nonverbal involvement is highly similarto immediacy, but the involvement constructis clearly more comprehensive at the oper-ational level than immediacy is. Further-more, Mehrabian (1969) emphasized thelink between the immediacy behaviors andinterpersonal attitudes. Thus, like the con-struct of intimacy, immediacy has been iden-tified with a single function. In general, in-creased involvement would be indicated bydecreased distance, increased gaze, touch,more direct body orientation, more forwardlean, greater facial expressiveness, longerspeech duration, more frequent or more in-tense interruptions, increased postural open-ness, more relational gestures, more frequenthead nods, and more intense paralinguisticcues. Again, modifications may be made inthis list, and the relative importance of thedifferent cues has yet to be determined.

Noninvolvement cues are those behaviorsthat are less salient to the dynamics of socialexchange. For example, Mehrabian foundthat rocking movements and leg or footmovements (when seated) were negativelyrelated to relaxation, whereas object manip-ulations and arm-position symmetry werepositively related to distress (Mehrabian,1970; Mehrabian & Ksionzky, 1972). Sim-ilarly, one class of self-manipulations, termedself-adaptors by Ekman and Friesen (1969),are viewed as being basically outside the in-teractive process. Instead, these behaviors,including rubbing, scratching, licking one'slips, or covering the eyes, have their originin serving personal needs (Ekman & Friesen,1969). It should be emphasized that suchnoninvolvement behaviors can still be im-portant and informative cues. In fact, suchbehaviors may provide useful informationabout personal states or traits. However,their direct influence on the interaction pro-cess is probably less than that of the involve-ment behaviors. In summary, it is proposedthat a behaviorally specified construct—non-verbal involvement—can be identified asconceptually distinct from the more abstractfunctional constructs relating to nonverbalexchange.

Functional Classification

The functional classification proposed herewill build on classifications offered by others(Argyle, 1972; Ekman & Friesen, 1969), butsome new distinctions will also be suggested.The purpose of the present classification isto provide a relatively comprehensive set offunctions that can be related to the man-agement of the involvement behaviors in so-cial exchange. The specific functional cate-gories proposed here are those of (a)

1 Relational gestures include those hand and/or armmovements that refer to the listener or the relationshipbetween the speaker and listener, for example, a handmovement by the speaker going out toward the listenerand back that might occur simultaneously with a com-ment such as "the two of us."

2 Most researchers distinguish between paralinguisticcues defining the vocal aspects of speech and the contentor meaning of the verbalization. The former is desig-nated as nonverbal (i.e., does not relate to the contentof speech), whereas the latter is clearly verbal.

Page 4: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

234 MILES L. PATTERSON

providing information, (b) regulating inter-action, (c) expressing intimacy, (d) exercis-ing social control, and (e) facilitating serviceor task goals. The first two functions areespecially useful in identifying the immedi-ate utility of isolated behaviors (e.g., a spe-cific touch, smile, or head nod). Thus, eachbehavior might be seen as serving either aninformational or regulatory function. Forunderstanding the patterns of behavior overtime, the last three functions—expressingintimacy, social control, and service-taskfunctions—may be more important. In ef-fect, the molecular functions of providinginformation and regulating interaction areindependent of the molar functions of inti-macy, social control, and service-task man-agement. In this way a given behavior canbe either informational or regulatory and,at the same time, be part of an overall pat-tern serving intimacy, social control, or ser-vice-task functions (Knowles, Note 3). Thefollowing section will describe briefly thesefunctions and their distinguishing charac-teristics.

Providing Information

Providing information might be describedas the most basic function of nonverbal be-havior. That is, in some sense everything anactor or encoder does is potentially infor-mative to the observer or decoder. Beyondthat description there is considerable dis-agreement in conceptual analyses of infor-mative nonverbal behavior. However, oneconceptual distinction that has considerableutility is a dichotomy between informativebehavior that is communicative and thatwhich is indicative (Ekman & Friesen, 1969;MacKay, 1972; Wiener, Devoe, Rubinow,& Geller, 1972). Such a distinction requiresan analysis of what the encoder, rather thanthe decoder, does in an interaction.3

For the purpose of developing the com-munication-indication contrast in the pres-ent theory, MacKay's (1972) emphasis onthe purposive nature of communication seemsmost useful. Specifically, MacKay claimsthat communicative behavior is purposivewhereas indicative behavior is not. Further-more, a communicative pattern would becharacterized by an evaluation of whetherthe behavior achieved its intended purpose.

Operationally, this would be manifested bychanges in the behavioral pattern followingsome possible criterion action. For example,I might use the behaviors of glancing at mywatch, fidgeting in my chair, and checkingmy appointment calendar to communicatethat I wanted to end an interaction. If thosebehaviors abruptly stopped when a visitorstarted to leave, that would p$vide evidencefor those behaviors being purposive, that is,communicative. If those behaviors continuedeven when the visitor had left, such behaviorswould be likely to be judged as indicative.That is, those behaviors were not affectedby the visitor's departure. Instead, they seemto indicate something about either a char-acteristic trait or a temporary state. A sim-ilar perspective was advanced by Ekman andFriesen (1969) in terms of the intention ofthe encoder. In their view a behavior is com-municative when a conscious intention canbe identified. The practical utility in consid-ering the distinction between communica-tion and indication lies in an attributionalevaluation of others' behavior, especiallywhen such behavior may be seen as sociallydesirable or normative (Jones & Davis,1965). In general, we are more likely to infera correspondent dispositional cause for somebehavior if we judge it as spontaneous orindicative than if we judge it as managed,purposive, or communicative.

Regulating Interaction

The second function, and probably the onethat is most automatic and least reflective,is that of regulating interaction. Before dis-cussing this function it is useful to draw acontrast between two basic types of socialsettings. Specifically, Goffman (1963, p. 83)has proposed that our naive or commonsensenotion of an interaction—two or more peopledirectly communicating, or at least snaringthe license to communicate, to the exclusion

3 A systems approach to describing communication(e.g., Watzlawick & Beavin, 1967) rejects the utilityof distinguishing between communication and indica-tion. More basically, this approach does not even acceptthe validity of the encoder-decoder contrast in analyzingcommunication. Instead, the systems approach holdsthat all social behavior is communication. Segmentingcommunication into encoder and decoder elements mis-represents the basic interpersonal nature of that process.

Page 5: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 235

of others present—be termed a focused in-teraction. In contrast, when individuals arenot directly communicating, but merelysharing a common presence, they can bedescribed as engaged in unfocused interac-tion.

In focused interaction the behavioralframework for interaction appears to be de-termined substantially by its "standing fea-tures," which include distance, body orien-tation, and posture (Argyle & Kendon,1967). As standing features these behaviorsremain relatively stable over the course ofan interaction. In general, the standing fea-tures must permit some minimum level ofinvolvement to facilitate a comfortable con-versation. In addition, the standing featuresset some rough limits on the opportunity forinitiating varying degrees of involvementthrough other nonverbal behaviors. For ex-ample, greater distances in interaction maketouching impossible.

In contrast to the standing features, dy-namic features such as gaze, facial expres-sion, and verbal intimacy affect the momen-tary changes in conversational sequences(Argyle & Kendon, 1967). Duncan (1972)has identified a variety of speaker paralin-guistic cues such as a change in the pitch ofthe voice, drawl on a final syllable, and de-creased loudness as related to smooth con-versational sequencing. In addition, listenerresponses such as smiling, positive headnods, or brief verbal reactions such as "mm-hmm," "yeah," or "I see" may serve not onlyto indicate agreement but also to signal thespeaker to continue or elaborate. Of course,the timing of these cues in conversation, theirintensity, and the extent to which they ap-pear in isolation or in combination with oneanother affects their influence on the se-quencing process. Comprehensive reviews ofthe role of nonverbal behavior in conversa-tional sequencing by Rosenfeld (1978) andFeldstein and Welkowitz (1978) can be con-sulted for the details that are beyond thescope of this paper.

In unfocused interactions the common ex-pectancy is that a low level of involvementwill not be exceeded. As in focused inter-actions, the standing features are primaryin controlling the level of involvement. If aperson wanted to maintain the interaction

as an unfocused one (i.e., avoid conversa-tion) in a setting like a library or waitingroom, a remote position facing away fromothers might be selected. With the exceptionof gaze avoidance and muted facial expres-siveness, the dynamic cues will obviously notapply to maintaining noninvolvement in un-focused interactions because they relate tospeech.

Discrete changes in nonverbal involve-ment at turn-taking points in conversationsmay often represent the function of regulat-ing interaction. For example, increased gazeand changes in facial expressiveness may bemore likely for both the speaker and listenerat the end of a conversational turn. In ad-dition, the listener, in anticipation of takingthe speaker role, may lean forward, assumea more tense or "ready" position, and evenbegin to gesture. Such changes, specific toturn-taking, may reflect more about thestructure or form of the interaction thanabout its explicit meaning. Although theunderlying purpose or function of the inter-action is not likely to be reflected in suchregulatory changes, the absence of such reg-ulation signals in conversations may bemeaningful and can be disruptive in an in-teraction. Practically, the function of regu-lating interaction sets some real lower limitsto the level of involvement in an exchange.That is, in focused interactions some mini-mal levels of proximity, gaze, directness ofbody orientation, and facial expressivenessare required for a comfortable interaction.

Expressing Intimacy

Much of the research on nonverbal in-volvement in interaction has been developedfrom and interpreted in an intimacy per-spective, that is, varied nonverbal involve-ment reflects the differential intimacy to-ward another. In general, intimacy might bedescribed as a bipolar dimension reflectingthe degree of union with or openness towardanother person. Practically, increased inti-macy is the result of greater liking or lovefor another or greater interest in or com-mitment to such a person. Generally, highintimacy is reflected in high levels of non-verbal involvement, but high intimacy is notalways manifested by high nonverbal in-

Page 6: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

236 MILES L. PATTERSON

volvement. Conversely, high nonverbal in-volvement may indicate something otherthan high intimacy. This latter circumstancewill be considered in greater detail in dis-cussing the next two functions. Represen-tative of the typical link between intimacyand involvement is Rubin's (1970) findingthat couples who score higher on a romanticlove scale (one aspect of intimacy) spentmore time in mutual gaze than did thosescoring lower. Similarly, in a review of spa-tial behavior, Patterson (1978b) reportedthat increased liking or attraction for an-other person resulted in closer approachesfor female-female and male-female pairs.

The intimacy function focuses on an af-fectively based reaction toward another per-son. The nonverbal involvement initiated asa result of that affective reaction is probablymore spontaneous than the involvement me-diated by the social-control and service-taskfunctions. Consequently, involvement pat-terns produced by differential intimacy to-ward another should generally not requirea self-conscious, managed routine. Withinthe constraints of the setting, differential in-volvement should flow relatively sponta-neously from the perceived level of intimacy,that is, there is a consistency between theunderlying affect toward another person(from very positive to very negative) and thelevel of nonverbal involvement. In otherwords, the actor's behavioral involvementwill typically indicate his/her intimacy withthe other person.

Social Control

The social-control function is a classifi-cation that has not received much emphasisin previous discussions of nonverbal behav-ior. One exception has been Henley's (1973,1977) work on the nonverbal correlates ofstatus or power differences developing fromsex role bias. Social control may be describedas involving some goal of exercising influ-ence to change the behavior of others. Morespecifically, the social-control process willfocus on producing reactions counter tothose expected without such influence. Thismay occur directly in trying to persuade oth-ers of our own particular viewpoint. For ex-ample, by implementing a moderately close

approach, increased gaze, and appropriateparalinguistic emphasis a particular verbalmessage may be more persuasive. Thus thecommunication will attempt to move the lis-tener from a position of opposition to one ofsupport. Another example is the use of gazeor touch to institute or reinforce status dif-ferences between individuals (Henley, 1973).In such a case the touch or gaze by the actorcan serve to reduce the status or freedom ofthe target person.

A less direct means of social control is theuse of any one of a variety of self-presen-tation patterns designed to have others viewan actor more favorably (Goffman, 1967,1972). Thus, a smiling, attentive expression,complemented by a forward lean, might bedesigned to show interest and create a fa-vorable impression in the other interactant.Such a behavioral strategy might be initiatedby an applicant in an employment interviewor might be used by a salesman to facilitateselling a product.

Nonverbal involvement serving a social-control function, compared to that servingan intimacy function, will be less sponta-neous and more self-conscious and managed.The social-control function will usually in-volve some evaluation of the intended pur-pose of the behavioral routine, that is, thebehavior continues or intensifies until itachieves its intended purpose. Thus the con-trast between the intimacy and social-con-jtrol functions can be clarified. Because theinvolvement mediated by the social-controlfunction is managed and purposive, it wouldtypically be classified as communicative be-havior. In contrast, the spontaneous involve-ment reflecting the intimacy function wouldtypically be classified as indicative behavior.

Services-Task

The final category, a service or task func-tion, identifies bases for nonverbal involve-ment that are essentially impersonal. Thatis, the particular level of involvement doesnot reflect anything about a social relation-ship between the individuals but only a ser-vice or task relationship. Heslin (Note 4)describes this type of function with respectto the use of touch. Examples of service re-lationships in which touch is appropriate are

Page 7: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 237

those of physician-patient, golf profes-sional-student, fireman-fire victim, and bar-ber/hairdresser-customer (Heslin, Note 4).Although high levels of nonverbal involve-ment are common in these and in some otherprofessional relationships, such involvementis merely the means to an end of treating,teaching, or otherwise serving the needs ofan individual. In many of these service ex-changes, the behavior of both participantscould be seen as following an ordered, pre-dictable sequence or script (Abelson, 1981).

Other occasions of variable involvementappropriate to this function may be lessstructured by role relationships and includea variety of exchanges relating to some com-mon tasks, for example, standing or sittingin a close arrangement so that materials canbe shared. In addition, this category mayinclude those adjustments necessary becauseof some sensory handicap, for example, a

i person with a hearing loss may have to sitcloser to others and try to gaze more directlyat them in order to understand conversationsbetter. Solitary tasks or activities that haveto be performed in the presence of othersmight be facilitated by low involvement.Reading, studying, or writing are examplesof activities in unfocused interactions thatoften require decreased involvement.

The service-task function might be dif-ferentiated from the intimacy and social-control functions in terms of the behavioralexchanges being more routine and havingless interpersonal relevance for the interac-tants. The routine nature of the service-taskexchanges is often reflected in their followinga particular script, for example, our servicecontacts with physicians or dentists or eventhe manner in which we relate to others insharing various work activities. Because in-terpersonal involvement in these exchangesis constrained by the norms of the setting,dispositional attributions for the other per-son's nonverbal behavior should be less fre-quent with the service-task function thanwith the intimacy and social-control func-tions.

The Sequential Model

The functional classification presentedhere provides a structure for analyzing non-

verbal involvement. However, this classifi-cation system by itself does not address theissue of the variety in nonverbal involvementrevealed across people and situations. Toappreciate and understand that variety, at-tention must be directed to the sequentialdevelopment of events prior to and duringthe interaction. The course of these sequen-tial processes and their relationship to thefunction of the interaction determine thepatterns of nonverbal exchange. The specificstages of the proposed model and their com-ponent elements are outlined in Figure 1.The discussion of this model will begin withthe antecedent factors and then proceedthrough the sequentially linked stages.

Antecedent Factors

The antecedent factors identified in Fig-ure 1 include personal, experiential, and re-lational-situational determinants of nonver-bal involvement. I will make no attempt toreview the extensive research on the rela-tionships between each of these classes ofvariables and nonverbal involvement. How-ever, I can briefly describe a few examplesfor each of the antecedent factors. First, thepersonal factors include those characteristicsof the individual or his/her group that con-tribute to contrasting patterns of nonverbalinvolvement. The personal factors that seemmost important are culture, sex differences,and personality. With respect to culture,Hall (1966) reports that Arab, southernMediterranean, and Latin American peopleexhibit higher levels of involvement than doBritish and northern European people. Sexdifferences are reflected in females mani-festing greater involvement than males in thefollowing behaviors: (a) closer interactiondistances (Aiello & Aiello, 1974; Patterson& Schaeffer, 1977), (b) higher levels of gaze(Dabbs, Evans, Hopper, & Purvis, 1980;Exline, 1963; Exline, Gray, & Schuette,1965; Libby, 1970), and (c) more frequenttouching (Jourard, 1966). The literature onpersonality correlates of nonverbal involve-ment is voluminous. However, it might benoted that several dimensions seem to bepotentially promising predictors of differ-ential nonverbal involvement, including so-cial approach-avoidance, internal-external

Page 8: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

238 MILES L. PATTERSON

locus of control, field dependence-indepen-dence, and self-monitoring (Patterson, inpress). Of course, additional personal factorssuch as age, socioeconomic class, occupation,or religion, among others, might also be re-lated to differential preferences for nonver-bal involvement, but there has been less re-search on those variables.

The experiential factors relate to the in-fluence of recent or similar experiences onlater interactions. Within some moderaterange of past interactive involvement, futurepreferences might be most influenced by thereinforcement consequences. That is, apleasant (or unpleasant) experience at amoderate level of involvement will increase(or decrease) the probability of seeking sim-ilar exchanges. As the levels of involvementbecome more extreme, homeostatic adjust-ments may begin to qualify the effect of theconsequences of past interaction. For ex-ample, "over-stimulation" (e.g., Milgram,1970) resulting from high levels of nonverbalinvolvement may result in avoidance of po-tentially involving interactions. A comple-mentary, physiologically based perspective

proposes that there is a reliable periodicityin the timing and intensity of social inter-action (Chappie, 1970; Hayes & Cobb, Note5). That is, like other biorhythms, the pref-erence for social activity follows a cyclic pat-tern. In contrast, an adaptation-level (Hel-son, 1964) orientation will suggest thatdeviations from the adaptation level (AL)of involvement are incorporated into a newAL, making the same "objective levels" ofinvolvement appear less extreme. Regardlessof the specific processes operating here, itseems safe to propose that past, relevant non-verbal involvement affects both the initiationand evaluation of future involvement. Again,it is assumed that the more recent and moresimilar experiences (relative to the targetexchange) are more influential.

The relational-situational factors aregrouped in a common category because theyoften interact with one another in specifyinga particular influence on nonverbal involve-ment. That is, the influence of the type ofrelationship between individuals is oftenmoderated by the nature of the setting. Infact, although there is considerable research

ANTECEDENTS PRE-|NTERACTION MEDIATORS INTERACTION PHASE

Figure 1. A sequential functional model of nonverbal exchange.

Page 9: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 239

on relationships and nonverbal involvement,there is relatively little on the influence ofthe situation on nonverbal involvement. Rep-resentative of the former is the trend forthose in more intimate relationships to standat closer distances than do those in less in-timate relationships (Patterson, 1978b).Similarly, increased liking (Kendon & Cook,1969) or increased love (Rubin, 1970) arerelated to increased duration of gazing. Thefrequency and intimacy of touch also seemsto reflect the degree of relationship betweeninteractants (Jourard, 1966).

One useful perspective for describing andevaluating the influence of the situation isthe behavior-setting approach of ecologicalpsychology (Barker, 1968; Wicker, 1979).This approach emphasizes the influence ofthe physical, social, and selection character-istics of a setting in contributing to the ho-mogeneity of behavior across individuals. Asetting like a church service or a businessoffice may determine involvement behaviormore strongly than will a setting such as aparty or a barbecue. In the former settingsthe impact of relational influences is usuallydiminished by the setting's norms and phys-ical design. In the party or barbecue setting,the social norms and physical design may bemore variable and, consequently, the rela-tional influences become more important.

Preinteraction Mediators

The effects of the antecedent factors onthe course of an impending interaction areassumed to be mediated by more covert pro-cesses. Thus, at the time an interaction isinitiated, specific mediators already providesome structure to the subsequent exchange.The first mediator, behavioral predisposi-tions, could be viewed as relatively stablecharacteristics of individuals. It is assumedthat these predispositions are usually moreor less automatic and often not well repre-sented cognitively. For example, behavioralpredispositions may commonly mediate theinfluence of the personal factors on nonver-bal involvement. That is, the involvementdifferences described earlier as a function ofculture, sex, or personality apparently rep-resent relatively stable preferences requiringlittle or no cognitive reflection. For example,

the tendency for females to exhibit higherlevels of nonverbal involvement than domales is a fairly general behavioral tendencythat probably operates outside of the indi-vidual's awareness. In effect, behavioral pre-dispositions describe a range of involvementone is likely to judge as comfortable or ap-propriate. One will be aware of that com-fortable range of involvement only to theextent that it is violated in an interaction.

A second mediator is that of potentialarousal change. The use of the term changein describing arousal is quite deliberate.Although it is common to expect increasedarousal following increased involvement, itis useful to consider decreases, as well asincreases, in arousal as critical events in non-verbal exchange (Patterson, 1978a). Specif-ically, when an individual is distressed orfearful, high nonverbal involvement may ac-tually facilitate decreased arousal and thedevelopment of more positive feeling states.In fact, in one recent study conducted onhospital patients, contrasting arousal, be-havioral, and affective reactions to a nurse'stouch were found as a function of the sex ofthe individual (Whitcher & Fisher, 1979).The female patients reacted to the touchwith decreased arousal and more pos-itive behavioral and judgmental reactions,whereas males reacted with increased arousaland more negative behavioral and judgmen-tal reactions.

It is also appropriate to qualify the arousal-change process with the term potential be-cause the role of arousal change in nonverbalexchange is probably not as pervasive as sug-gested in the arousal model (Patterson,1976). Although the Whitcher and Fisherstudy (1979) provides strong support for themediating role of arousal change, our ownresearch has not consistently demonstratedsuch an effect (Patterson et al., 1981; Pat-terson, Roth, & Schenk, 1979). At the sametime it seems important to recognize evena tentative role for arousal change becauseit may affect the pattern of behavioral ad-justments in two different ways. First,changes in arousal might motivate behav-ioral adjustment or differentially facilitatedominant adjustment patterns (Zajonc,1965). Second, arousal change may serve asa signal to initiate a meaning analysis of the

Page 10: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

240 MILES L. PATTERSON

changing circumstances (Mandler, 1975,chapter 4). Such an analysis is one form ofthe last mediator: cognitive-affective assess-ment.

Cognitive-affective assessment is intendedto reflect the initiation of any kind of eval-uation, from the very simple to the very com-plex, for example, from a dichotomous judg-ment of good or bad to activating a schema(Markus, 1977) or script (Abelson, 1981).Zajonc (1980) claims that simple affectivereactions or preferences are very rapid, pri-mary responses to. stimuli. Such affectiveresponses can be differentiated from moreneutral cognitions about the same stimuli.For example, in meeting a stranger globalaffective responses or first impressions candevelop in a fraction of a second (Zajonc,1980). Similar but more specific affectivereactions to the behavior of others after theinteraction has started may leave us with apleasant or unpleasant feeling and no aware-ness of the source of those affects. Obviously,such affective reactions can set the tone foran exchange and stimulate an increase incognitive activity. More elaborate expectan-cies often play a role in the unfolding of aninteraction. Darley and Fazio (1980) pro-pose a variety of processes relating to boththe actor (perceiver) and the target personin an interaction that facilitates the confir-mation of expectancies.

Interaction Phase

The preinteraction mediators serve tolimit an individual's preferred involvementlevel and sensitize him/her to functionaljudgments about the interaction. To the ex-tent that specific functional expectancies areactivated, the initial behavioral involvementwill tend to be more structured. For example,if a friend introduces me to a potential em-ployer, my initial behavior will be affectedby a social-control perception of the inter-action. That is, I will be trying to behave ina way that might induce a favorable reac-tion, for example, I might smile a little morethan usual, lean forward a bit, maintain ahigh level of gaze, and nod my head occa-sionally. If that friend were to introduce thesame person without identifying him in anyway, my habitual involvement pattern (be-havioral predisposition) should contribute

more to determining my initial involvement.In turn, a perceived function of that inter-action may only develop later as a result ofattributions precipitated by the nonverbalexchange. Thus, in the first case explicitfunctional expectancies guide the initiationof nonverbal involvement. In the second casebehavioral predispositions are more impor-tant in determining initial involvement, andfunctional expectancies may be inferredlater from the course of the behavioral ex-change.

Obviously, an exchange is guided not onlyby each person's individually determined in-volvement but also by an assessment of whatis appropriate behavior from the other per-son. It is assumed that the converging influ-ences of behavioral predispositions and bothrelational and functional cognitions deter-mine a range of appropriate involvementexpected of the other person. This range orexpectation level of appropriate involvementbecomes a standard against which the otherperson's behavior can be judged (Cappella& Greene, Note 6). When the discrepancybetween the expected level of involvementand the other person's actual involvement isminimal, an exchange can be described asstable. When the discrepancy is too great,the exchange can be described as unstable.Involvement patterns in stable exchangesshould be less variable over time and requirelittle or no cognitive-arousal mediation.However, as Figure 1 indicates, unstableexchanges precipitate further cognitive-arousal mediation.

It seems fairly clear that being the recip-ient of inappropriately high levels of non-verbal involvement (i.e., involvement typi-cally exceeding the expected level) can affectarousal (Gale, Lucas, Nissim, & Harpham,1972; Kleinke & Pohlen, 1971; Nichols &Champness, 1971; Whitcher & Fisher,1979). Again, as mentioned earlier, arousalchange may trigger some cognitive-affectiveassessment or the assessment of the otherperson's involvement may precipitate arousalchange. It might be noted that Cognitive-Affective Assessment II (see Figure 1) de-picted in the interaction phase, like Cogni-tive-Affective Assessment I, is intended tobe broader than the self-labeling focus of thearousal model (Patterson, 1976). In fact,

Page 11: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 241

attributions about the other person and his/her motivation may have greater utility thanthose focused on one's own feelings (Ells-worth, Note 1).

The cognitive-arousal processes can leadeither to a simple direct adjustment in one'sinvolvement level or to a more pervasive as-sessment of the interaction's function. In thelatter case the resulting adjustment willprobably follow from a reassessment of theperceived function. If these adjustments areadequate, as measured by a new discrepancyanalysis of the partner's subsequent behav-ior, then the exchange will tend to stabilize.If the discrepancy continues, a continuedstate of instability is signaled and anotheradjustment cycle will be initiated. Should aninteraction continue to be unstable and un-affected by the adjustments of one or bothparties, an earlier than normal terminationwould be expected. An example of this in anunfocused interaction is the finding in Felipeand Sommer's (1966) study on spatial in-vasions. Those who were approached mostclosely (6 inches or less, side-by-side) left thesetting much sooner than did either the con-trols seated alone or those who were ap-proached at less extreme distances.

Finally, the termination of an interactionleaves a residue of cognitive and affectivereactions that feed back to influence the an-tecedent factors. Of course, with respect tothe personal antecedents, such feedback isunlikely to change one's age, race, or sex,but that feedback may alter interpersonalattitudes or values associated with thosecharacteristics. Further, changes in one'spersonality, though probably not great afterany single exchange, are certainly possibleover time. The experiential factor, by defi-nition, includes residual influences from re-cent or important previous exchanges. Forrepeated or similar exchanges, the qualityof the previous exchange can often deter-mine the direction of the impending inter-action. Finally, the outcome of a given ex-change influences complementary self-,other-, and setting-selection processes de-scribed in the relational-situational factor.Depending on the valence of the outcome,distinct approach or avoidance tendenciesregarding the other person or the setting maydevelop.

Summarizing the Model

The sequential functional model assumesthat each person brings to an interaction aspecific set of personal characteristics, pastexperiences, and relational-situational con-straints. This group of antecedent factorsexercises its influence covertly at the prein-teraction stage by (a) determining behav-ioral predispositions for differential involve-ment, (b) precipitating arousal change, and(c) developing cognitive and affective expec-tancies. The joint effect of these three me-diators shapes the perception of the inter-action's function and limits the range ofinvolvement initiated. The combined influ-ence of each person's preferred involvementlevel and functional-relational cognitions, inturn, shapes expectancies about the otherperson's level of involvement. A stable ex-change is defined as one in which the dis-crepancy between the expected and actualinvolvement of the other person is minimal.When that discrepancy is large, a conditionof instability results, triggering cognitive-arousal processes. These processes lead toadjustments in nonverbal involvement thatmay or may not be accompanied by a reas-sessment of the interaction's function. Suc-cessful involvement adjustments and amore complementary functional reassessment(relative to the other person's functional as-sessment) should facilitate a more stableexchange.

Measuring Arousal and CognitiveResponses

In order to test the model, some reliablemeans of monitoring arousal and cognitiveresponses must be available. Fortunately,that does seem to be the case. In two dif-ferent studies we have used electrodermalmeasures in examining the effects of con-federate involvement level on arousal. Spe-cifically, skin conductance level (SCL) andskin conductance response (SCR) measureswere monitored in one experiment utilizinga bogus waiting period (Patterson et al.,1981) and in another experiment utilizinga more structured interaction (Ickes, Pat-terson, Rajecki, & Tanford, in press). Heartrate has also been used as an arousal indexin another study examining reactions to a

Page 12: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

242 MILES L. PATTERSON

confederate's programmed changes in non-verbal involvement (Coutts, Schneider, &Montgomery, 1980). The specific results ofthese studies will be discussed in some detaillater. For now it should be noted that theyprovide examples of measures amenable tothe laboratory setting. Reactivity of the sub-jects is a problem in such research, but theuse of engaging tasks or bogus waiting pe-riods can help to minimize this problem.Detailed information about the use of psy-chophysiological measures might be ob-tained from a recent text by McGuigan(1978).

Assessing the mediating cognitions pre-sents some obvious procedural difficulties. Acommonly used technique involves havingsubjects complete various ratings on the in-teraction, the other person, and themselvesafter the interaction has ended. It is gener-ally assumed that such ratings reflectthoughts and feelings present during the in-teraction. Although such ratings may be in-formative, they cannot provide much insightinto the sequential changes in thoughts andfeelings, especially as they may relate to spe-cific critical changes in the patterns of non-verbal behavior. However, there are meansby which the assessment of sequentialchanges in thoughts could be accomplished.

First, it should be noted that the simplelisting of thoughts for a specific intervalcould be managed quite easily. For example,Caccioppo and Petty's (1981b) thought-list-ing procedure requires sugjects to record theseparate thoughts they recalled from a spec-ified period in the experiment. Caccioppoand Petty report that this measure can bescored reliably, is sensitive to experimentalmanipulations, and can reflect motivationalchanges. In work in our own laboratory, weare employing a variant of the thought-list-ing procedure with the help of a videotapereplay of the interaction. The purpose of thistechnique is simply to assess changing cog-nitions over the period of the interaction.With this procedure a subject simply stopsthe videotape at the point where he/she re-calls a specific thought or feeling presentduring the interaction. The experimenterrecords the time of that thought as the sub-ject writes it on a response sheet. Subjects

usually recall one to three thoughts per min-ute and seem to be quite candid in their re-ports. We have been able to score these re-sponses reliably in terms of their focus (self,other, self-other, environment) and valence.

A second means of assessing the focus ofthoughts is the analysis of forms of personalpronoun use. Wegner and Giuliano (1980)used a type of sentence completion test re-quiring subjects to choose among differentpronouns. They found that subjects whosearousal was increased by running in placechose more first-person singular pronounsthan did subjects in a control condition. Wehave been working with a similar classifi-cation of pronoun use in the context of anongoing interaction. Initial results show thatpersonality characteristics are related to dif-ferent patterns of pronoun use. For example,self-monitoring was negatively correlatedwith frequency of first-person singular pro-noun use, and inner directedness was nega-tively correlated with frequency of third-per-son pronoun use (Reidhead, Ickes, &Patterson, Note 7). By taking this measurefrom the videotape record of interactions,the specific time of usage can be identifiedand related to discrete arousal and behav-ioral changes.

Finally, in situations where conversationis either impossible (e.g., a subject waitingalone) or proscribed, a psychophysiologicalmeasure, the electromyogram, may be par-ticularly useful. The electromyogram (EMG)measures the degree of muscle activity mon-itored though electrodes placed at specificlocations on the body. In a recent articleCaccioppo and Petty (198la) reviewed theresearch on EMG records of facial musclescontrolling expressive movements in theface. Of particular interest are results re-lated to changes in facial muscle activityover intervals in which cognitive activityshould be maximized, for example, followingpersuasive messages. In general, Caccioppoand Petty found increased speech-muscleactivity following the forewarning of per-suasive messages and during the actualpresentation of the messages. Further, it ap-pears that the anticipation of positive versusnegative events activates different patternsof facial muscles. This kind of measure

Page 13: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 243

might be employed to supplement self-re-ports of thoughts from subjects who arewaiting to meet another person, for example,following an expectancy manipulation.

In summary, the more subtle and elusivearousal and cognitive responses, which arecritical mediators in this theory, can be mea-sured. In particular, recently developed tech-niques of cognitive assessment provide thepotential for measuring some aspects of rel-atively continuous thought patterns in inter-actions. With the exception of the unobtru-sive classification of pronoun usage, theother measures described can be quite re-active. Sensitivity to this concern and cre-ativity in .structuring the interaction settingcan make the problem of reactivity a man-ageable one.

Empirical Evidence

One index of the viability of this sequen-tial model is the degree to which results in-consistent with the predictions of other the-ories (e.g., equilibrium theory and the arousalmodel) might be explained by this model. Infact, there is a group of studies whose resultsare quite compatible with the predictions ofthis new model. In particular, these findingssupport the specific contrast between the in-timacy and social-control functions of non-verbal involvement and the role of differ-ential expectancies in determining thosefunctions.

The first set of results comes from one ofour studies testing the earlier arousal model.In that study we examined changes in themale subjects' arousal and behavioral pat-terns as a function of programmed involve-ment changes by a male confederate (Pat-terson et al., 1981). This took place duringa bogus waiting period before an experi-mental task on "physiological correlates ofproblem solving." The subject and confed-erate were seated adjacently, both orientedtoward the table on which the problem solv-ing materials were placed. During the firstfew minutes of the waiting period, baselinesfor the electrodermal measures of arousalwere established. After that the confederateinitiated a conversation and then alternatedintervals of high and low involvement (within-

subject factor) over the remainder of thewaiting period. Three different orders, onecontrol and two experimental sequences,were investigated.

A hypothesized increase in arousal wasfound for the lean-touch manipulation butonly when it occurred in the middle, ratherthan at the start, of the confederate's con-versation. Related to that increased arousalwere tendencies for increased gaze towardthe confederate and increased talking. Theorder-specific effects of the lean-touch ma-nipulation suggest that such behavior mayserve different functions at different pointsin a conversation. Leaning toward andtouching another person at the start of con-versation probably serves a simple functionof regulating interaction. That is, given thenonconfronting orientations of the subjectand confederate, lean and touch become aconvenient way to get the other person's at-tention and make a comment. In that contextlean and touch fall within normative expec-tancies for involvement. In contrast, thesame behavior in the middle of a conversa-tion is likely to be outside the range of nor-mative expectancies, especially in an inter-action between males. If the two instancesof lean and touch do represent different func-tions, the resulting contrast in their arousaland behavioral consequences are consistentwith the distinctions made by the sequentialmodel.

The remaining studies all deal with theeffects of various expectancies on subsequentchanges in nonverbal involvement. Thearousal model predicts that negative expec-tancies should lead to compensatory adjust-ments, whereas positive expectancies shouldlead to reciprocal adjustments. One studytested this very prediction by manipulatingpositive or negative feedback from the con-federate prior to a critical interaction (Couttset al., 1980). Following a baseline period inthat interaction, the confederate either heldnonverbal involvement constant or increasednonverbal involvement from its baselinelevel. Increased involvement by the confed-erate, in the form of increased gaze and amore direct body orientation, did produce anincrease in a heart-rate measure of arousal.However, the behavioral adjustment pat-

Page 14: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

244 MILES L. PATTERSON

terns were not consistent with the predictionsof the arousal model.

Specifically, the subjects in the positivefeedback—increased involvement conditionshowed no significant changes in nonverbalinvolvement. Even more surprising was thepattern of reciprocated involvement fromsubjects in the negative feedback—increasedinvolvement condition—just the opposite ofthat predicted by the arousal model. Cor-relational analyses showed generally positiverelationships between involvement changeand rated liking of the confederate in thepositive feedback condition. That is, the rel-ative change in the subject's involvement waspositively correlated with the rated liking ofthe confederate, a finding consistent with anintimacy function. However, in the negativefeedback condition (showing significant re-ciprocal adjustments), the increased involve-ment was not generally related to rated lik-ing of the confederate. The negative-feedbacksubjects' pattern of reciprocation cannot beexplained by an intimacy or liking function,but it is thoroughly consistent with a social-control function. In fact, Coutts et al. (1980)suggest an interpretation for these resultsthat clearly describes a social-control func-tion: "Under these circumstances, the neg-ative condition subjects may have viewed theaccomplice's subsequent increase in imme-diacy as a return to a more appropriate levelof intimacy and thus exaggerated their ownnonverbal immediacy to encourage if (p.588, italics added).

The results from other recent researchshed additional light on the relationship be-tween expectancies and functions of nonver-bal involvement. Two experiments were con-ducted to examine the effects of expectancieson subsequent patterns of nonverbal ex-change (Ickes et al., in press).4 Consistentresults were found across two differing pro-cedures in these experiments. Specifically,in Experiment 1 both friendly and unfriendlypreinteraction expectancies led subjects toinitiate greater nonverbal involvement com-pared to a no-expectancy control. Most strik-ing was the increased smiling in the, un-friendly condition compared to both thecontrol and friendly conditions. Althoughthe behavioral patterns of the friendly andunfriendly conditions were otherwise gen-

erally similar, their contrasting patterns ofrated impressions of their partners suggestthat differing underlying strategies may beoperating. In particular, the friendly condi-tion was characterized by a high level ofattraction, but the unfriendly condition wascharacterized by a low level of attraction anddistrust of the partner's seemingly friendlybehavior.

In Experiment 2 the expectancy manipu-lation was structured in terms of the other'ssimilarity or dissimilarity on personalitymeasures, In the similar condition subjectswere told that the other person was probablythe type of individual they would choose fora friend and, consequently, the interactionshould be enjoyable. Contrasting informa-tion was given in the dissimilar condition.As in Experiment 1, subjects in the dissim-ilar condition smiled more and, at the sametime, rated their partners more negativelythan did subjects in the similar condition.Nonsignificant trends on the remaining be-havioral measures—gaze, duration of talk-ing, and body orientation—also indicatedgreater involvement for the dissimilar sub-jects. Furthermore, subjects in the dissimilarcondition, compared to those in the similarcondition, experienced increased arousal(measured by electrodermal responses) im-mediately after the expectancy manipula-tion. This last effect is consistent with thismodel's prediction of greater physiologicalarousal in anticipating unusual or uncom-fortable interactions.

Across the two experiments, unfriendly ornegative expectancies produced as high orhigher (smiling) involvement than didfriendly or positive expectancies, even thoughattraction to the partner was predictablyvery different as a function of the expec-tancy. The pattern present in the positiveexpectancy conditions is consistent with anattraction-mediated or intimacy-based func-

4 While three of the authors (Ickes, Rajecki, & Tan-ford) were conducting Experiment 1, the remaining au-thor (Patterson) was independently conducting Exper-iment 2. The two experiments differed in terms of theirspecific predictions, procedures, interaction context, andsubject populations but nevertheless were remarkablysimilar in terms of the dependent measures, the patternof obtained results, and their implication for expectan-cies and patterns of nonverbal involvement.

Page 15: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 245

tion. In contrast, the pattern present in thenegative expectancy conditions is consistentwith a social-control function. This generalpattern of increased involvement followinga negative expectancy has also been reportedby Bond(1972).5

Finally, it is interesting to note that thelocus of the major contrast between the ex-pectancy conditions—the face—is one thatis critical in attempting deception (i.e., smil-ing often increases in deceptive encounters;Ekman & Friesen, 1974; Mehrabian, 1971).That, too, is consistent with the interpreta-tion that the increased involvement in thenegative expectancy conditions reported hererepresented a managed or social-control per-formance.

The results of all of these studies suggestclearly that understanding and predictinginvolvement patterns in interaction requiresattention to more than just the intimacyfunction. In particular, the evidence pre-sented here (Bond, 1972; Coutts et al., 1980;Ickes et al., in press) shows that an intimacy-based function cannot explain high levels ofinvolvement resulting from the expectationof an unfriendly or unpleasant interaction.Such negative expectancies can producearousal and lead to exaggerated involve-ment. Those behavioral adjustments are ap-parently designed to make the partnermore responsive and the interaction morepleasant.

Discussion

This model of nonverbal exchange pro-poses that interaction behavior could be seenas the product of a sequence of related ev-ents. Perhaps the most basic feature of themodel is the distinction proposed betweenthe behaviors of interest in an exchange andthe functions served by them. Such a dis-tinction rests on the recognition that thesame behavioral patterns can serve very dif-ferent functions in an interaction. A func-tional analysis of interaction seems to haveconsiderable utility, whether one accepts theclassification offered here or some alternatesystem. The current emphasis in psychologyon cognitive approaches to understandingsocial behavior are obviously consistent witha functional model.

The proposed contrasts among functionsand their related arousal, cognitive, and be-havioral components distinguish this modelfrom previous models of nonverbal behavior.A critical issue here is determining the cir-cumstances that are likely to precipitate theinitiation of one or another function. In gen-eral, when an interaction is less structured,and less evaluative in nature, it is more likelythat such an interaction will reflect the in-timacy function. This should result in agreater role for both behavioral predisposi-tions and interpersonal affect in determiningone's own level of involvement and expec-tancies about the involvement of the otherperson. When an interaction is less sponta-neous, more structured, and more highlyevaluative in nature, it is more likely thatsuch an interaction will reflect the social-control function, that is, nonverbal involve-ment is managed to effect a change in theother person's behavior. This should resultin activating strategies or scripts that directan actor's behavior more than either behav-ioral predispositions or the other person'spreceding reactions do.

An example of a circumstance than caninitiate the social-control function is a par-ticularly negative expectancy about anotherperson. Such a negative expectancy can trig-ger a behavioral strategy designed to pro-duce a favorable response from that person(Bond, 1972; Coutts et al., 1980; Ickes et al.,in press). Ickes et al. speculate about theconditions that are likely to precipitate astrategy of managed involvement. Specifi-cally, the actor should view the other per-son's anticipated behavior as undesirable butbelieve that the anticipated behavior is mod-ifiable by adopting a behavioral pattern thatcontrasts with that anticipated behavior. Infact, the results from the Bond (1972) andIckes et al. (in press) studies showed that theactor's strategy of increased involvement did

5 Bond (1972) manipulated forewarnings about meet-ing a "cold" or "warm" person in an interaction. Al-though specific behavior contrasts were not reported,judges rated subjects who received the cold expectancyas acting more warmly in the target person's presencethan did subjects who received the warm expectancy.In addition, these warm-acting (but cold-expecting)subjects actually induced the target person to act morewarmly in return.

Page 16: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

246 MILES L. PATTERSON

positively affect the other person's behavior,that is, the other person smiled more in theinteraction. A complementary predictionwould be made in the contrasting circum-stance in which too much warmth or friend-liness is expected, that is, the actor shouldexaggerate his/her low involvement in anattempt to influence the other person to dothe same. An example of this latter circum-stance would be managed low involvementin response to anticipated, but unwanted,sexual intimacy. Such a "cold shoulder"strategy would be designed to discourage in-creased involvement and thereby permit amore comfortable exchange for the actor.

Some mention should also be made of thethird molar function, the service-task func-tion. When individuals share common ex-pectancies about their roles in various serviceor task relationships, their involvement pat-terns should be "scripted" by those expec-tancies. In that way differential involvementis ascribed to the requirements of the sharedactivity and not attributed to the other per-son. However, when common expectanciesare not shared, the involvement (either highor low) of the first person is more likely tobe discrepant from the second person's an-ticipated level of involvement. This conditionof instability can lead to misattributions bythe second person that focus on intimacy orsocial-control motives. Such misattributionswill, in turn, determine the pattern of non-verbal adjustment. Of course, the adjust-ments of the second person may be discrep-ant from the first person's service-taskexpectancies and the condition of instabilitywill continue. In general, to the extent thatfunctional attributions are inconsistent withone another, the potential for instability inthe exchange remains high.

Finally, it should be noted that results ina related area of research, verbal intimacyor self-disclosure, can be explained in termsof the functional distinctions proposed bythis model. The clearly predominant patternfound in that research is one of reciprocity,a pattern even stronger than the typical com-pensation pattern found in the research onnonverbal exchange (see Cappella, 1981,and Firestone, 1977, for discussions of thecontrasting verbal and nonverbal patterns).

Why does increasingly intimate self-disclo-sure from one person lead so consistently tocomparable disclosure from another? At-traction does not seem to be an adequatemediating explanation (Firestone, 1977) be-cause high disclosure from a deviant (unlik-able) other is reciprocated as much as highdisclosure from a liked other (Derlega, Har-ris, & Chaikin, 1973). To the extent thatattraction per se cannot explain this pattern,such verbal reciprocity is not likely to be inthe service of the intimacy function. Cap-pella (1981) suggests that a potentially im-portant factor in the reciprocity of verbaldisclosure is the pressure for topic contin-uation in a verbal exchange. That is, a com-fortable and reasonable verbal exchange re-quires related responses from the inter-actants. In many cases this may occur almostspontaneously, but some matching or rec-iprocity is likely to be the product of a man-aged social-control function designed to ei-ther influence the other person or simplymake the interaction more comfortable. Ifa relatively subtle behavior like smiling canbe used either in influencing the involvementbehavior of another person (Ickes et al., inpress) or in deception (Ekman & Friesen,1974; Mehrabian, 1971), then managingverbal behavior for similar purposes is clearlypossible. In fact, our common experiencesshould make that evident. Furthermore, tothe degree that involvement behaviors suchas gaze, orientation, lean, and others are lessmanageable than are verbal behavior andsmiling, they may be less likely to be usedin the service of a social-control function.Thus, initiating social-control motives shouldproduce a greater change (from baseline lev-els) in smiling and verbal behavior than inless "deliberate" involvement behaviors. Sucha prediction is one that is eminently testablein future research.

In conclusion, this sequential functionalmodel attempts to provide an integrative,comprehensive approach to understandingnonverbal exchange. The various processesand linkages in the model present opportu-nities for generating a number of testablehypotheses. Although this model is focusedspecifically on nonverbal exchange, it ishoped that the type of sequential framework

Page 17: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 247

proposed here may provide an importantperspective for all researchers examining so-cial behavior.

Reference Notes1. Ellsworth, P. C. Some questions about the role of

arousal in the interpretation of direct gaze. Paperpresented at the annual meeting of the AmericanPsychological Association, San Francisco, August1977.

2. Knowles, E. S. Affective and cognitive mediators ofspatial behavior. Paper presented at the annualmeeting of the American Psychological Association,San Francisco, August 1977.

3. Knowles, E. S. Personal communication, January1981.

4. Heslin, R. Steps toward a taxonomy of touching.Paper presented at the annual meeting of the Mid-western Psychological Association, Chicago, May1974.

5. Hayes, D. P., & Cobb, L. Infraradian periodicityin human social interaction. Paper presented at theannual meeting of the American Psychological As-sociation, New Orleans, September 1974.

6. Cappella, J. N., & Greene, J. V. A discrepancy-arousal explanation of mutual influence in expres-sive behavior for adult and infant-adult interaction.Unpublished manuscript, 1981. (Available fromJ. N. Cappella, Department of Communication Arts,Vilas Hall, University of Wisconsin, Madison, Wis-consin 53706.)

7. Reidhead, S., Ickes, W., & Patterson, M. L. Personalpronoun use as an indicator of personality. Manu-script in preparation, 1982.

ReferencesAbelson, R. P. Psychological status of the script concept.

American Psychologist, 1981, 36, 715-729.Aiello, J. R., & Aiello, T. D. The development of per-

sonal space: Proxemic behavior of children 6 through16. Human Ecology, 1974, 2, 177-189.

Argyle, M. Non-verbal communication in human socialinteraction. In R. A. Hinde (Ed.), Non-verbal com-munication. Cambridge, England: Cambridge Uni-versity Press, 1972.

Argyle, M., & Dean, J. Eye-contact, distance, and af-filiation. Sociometry, 1965, 28, 289-304.

Argyle, M., & Kendon, A. The experimental analysisof social performance. In L. Berkowitz (Ed.), Ad-vances in experimental social psychology. New York:Academic Press, 1967.

Argyle, M., Lalljee, M., & Cook, M. The effects ofvisibility on interaction in a dyad. Human Relations,1968, 21, 3-17.

Barker, R. G. Ecological psychology: Concepts andmethods for studying the environment of human be-havior. Stanford, Calif.: Stanford University Press,1968.

Bond, M. H. Effect of an impression set on subsequentbehavior. Journal of Personality and Social Psy-chology, 1972, 24, 301-305.

Breed, G. The effect of intimacy: Reciprocity or retreat?British Journal of Social and Clinical Psychology,1972, /;, 135-142.

Caccioppo, J. T., & Petty, R. E. Electromyograms asmeasures of extent and affectivity of information pro-cessing. American Psychologist, 1981, 36, 441-456.(a)

Caccioppo, J. T., & Petty, R. E. Social psychologicalprocedures for cognitive response assessment: Thethought listing technique. In T. V. Merluzzi, C. R.Glass, M. Genest (Eds.), Cognitive assessment. NewYork: Guilford Press, 1981. (b)

Cappella, J. N. Mutual influence in expressive behavior:Adult-adult and infant-adult dyadic interaction.Psychological Bulletin, 1981, 89, 101-132.

Chapman, A. J. Eye contact, physical proximity andlaughter: A reexamination of the equilibrium modelof social intimacy. Social Behavior and Personality,1975, 3, 143-155.

Chappie, E. D. Culture and biological man: Explora-tions in behavioral anthropology. New York: Holt,Rinehart & Winston, 1970.

Coutts, L. M., Schneider, F. W., & Montgomery, S.An investigation of the arousal model of interpersonalintimacy. Journal of Experimental Social Psychol-ogy, 1980, 16, 545-561.

Dabbs, J. M., Jr., Evans, M. S., Hopper, C. H., &Purvis, J. A. Self-monitors in conversation: What dothey monitor? Journal of Personality and SocialPsychology, 1980, 39, 278-284.

Darley, J. M., & Fazio, R. H. Expectancy confirmationprocesses arising in the social interaction sequence.American Psychologist, 1980, 35, 867-881.

Derlega, V. J., Harris, M. S., & Chaikin, A. L. Self-disclosure, liking and the deviant. Journal of Exper-imental Social Psychology, 1973, 9, 277-284.

Duncan, S., Jr. Some signals and rules for taking speak-ing turns in conversations. Journal of Personality andSocial Psychology, 1972, 23, 283-292.

Ekman, P., & Friesen, W. V. The repertoire of non-verbal behavior: Categories, origins, usage and cod-ings. Semiotica, 1969, /, 49-97.

Ekman, P., & Friesen, W. V. Detecting deception fromthe body or face. Journal of Personality and SocialPsychology, 1974, 29, 288-298.

Exline, R. V. Explorations in the process of person per-ception: Visual interaction in relation to competition,sex, and need for affiliation. Journal of Personality,1963, 31, 1-20.

Exline, R., Gray, D., & Schuette, D. Visual behaviorin a dyad as affected by interview content and sex ofrespondent. Journal of Personality and Social Psy-chology, 1965, /, 201-209.

Feldstein, S., & Welkowitz, J. A chronography of con-versation: In defense of an objective approach. InA. W. Siegman & S. Feldstein (Eds.), Nonverbalbehavior and communication. Hillsdale, N.J.: Erl-baum, 1978.

Felipe, N. J., & Sommer, R. Invasion of personal space.Social Problems, 1966, 14, 206-214.

Firestone, I. J. Reconciling verbal and nonverbal modelsof dyadic communication. Environmental Psychologyand Nonverbal Behavior, 1977, 2, 30-44.

Page 18: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

248 MILES L. PATTERSON

Foot, H. C., Chapman, A. J., & Smith, J. R, Friendshipand social responsiveness in boys and girls. Journalof Personality and Social Psychology, 1977,35,401-411.

Foot, H. C,, Smith, J. R., & Chapman, A. J. Individualdifferences in children's social responsiveness in hu-mour situations. In A. J. Chapman & H. C. Foot(Eds.), It's a funny thing, humour. Oxford, England:Pergamon Press, 1977.

Gale, A., Lucas, B., Nissim, R., & Harpham, B. SomeEEG correlates of face-to-face contact. British Jour-nal of Social and Clinical Psychology, 1972, / / , 326-332.

Goffman, E. Behavior in public places. New York: FreePress, 1963.

Goffman, E. Interaction ritual. Garden City, N.Y.:Anchor, 1967.

Goffman, E. Relations in public. New York: HarperColophon, 1972.

Hall, E. T. The hidden dimension. New York: Double-day, 1966.

Harrison, R. P. Nonverbal communication. In I. de SoloPool et al. (Eds.), Handbook of communication. Chi-cago: Rand McNally, 1973.

Helson, H. Adaptation-level theory. New York: Harper& Row, 1964.

Henley, N. M. Status and sex: Some touching obser-vations. Bulletin of the Psychonomic Society, 1973,2, 91-93.

Henley, N. M. Body politics: Power, sex, and nonverbalcommunication. Englewood Cliffs, N.J.: Prentice-Hall, 1977.

Ickes, W., Patterson, M. L., Rajecki, D. W., & Tanford,S. Behavioral and cognitive consequences of recip-rocal versus compensatory responses to pre-interac-tion expectancies. Social Cognition, in press.

Jones, E. E., & Davis, K. From acts to dispositions: Theattribution process in person perception. In L. Ber-kowitz (Ed.), Advances in experimental social psy-chology (Vol. 2). New York: Academic Press, 1965.

Jourard, S. M. An exploratory study of body-accessi-bility. British Journal of Social and Clinical Psy-chology, 1966, 5, 221-231.

Jourard, S. M., & Friedman, R. Experimenter-subject"distance" and self-disclosure. Journal of Personalityand Social Psychology, 1970, 15, 278-282.

Kendon, A. Some functions of gaze direction in socialinteraction. Acta Psychologies, 1967, 26, 22-63.

Kendon, A., & Cook, M. The consistency of gaze pat-terns in social interaction. British Journal of Psy-chology, 1969, 60, 481-494. .

Kleinke, C. L., & Pohlen, P. D. Affective and emotionalresponses as a function of other person's gaze andcooperativeness in a two-person game. Journal ofPersonality and Social Psychology, 1971, 17, SOS-SIS.

Libby, W. L. Eye contact and direction of looking asstable individual differences. Journal of Experimen-tal Research in Personality, 1970, 4, 303-312.

MacKay, D. M. Formal analysis of communicative pro-cesses. In R. A. Hinde (Ed.), Non-verbal commu-nication. Cambridge, England: Cambridge UniversityPress, 1972.

Mandler, G. Mind and emotion. New York: Wiley,1975.

Markus, H. Self-schemata and processing informationabout the self. Journal of Personality and SocialPsychology, 1977, 35, 63-78.

McGuigan, F. J. Cognitive psychophysiology: Princi-ples of covert behavior. Englewood Cliffs, N.J.: Pren-tice-Hall, 1978.

Mehrabian, A. Some referents and measures of non-verbal behavior. Behavior Research Methods andInstrumentation, 1969, /, 203-207.

Mehrabian, A, A semantic space for nonverbal behavior.Journal of Consulting and Clinical Psychology, 1970,35, 248-257.

Mehrabian, A. Nonverbal betrayal of feeling. Journalof Experimental Research in Personality, 1971, 5,64-73.

Mehrabian, A., & Ksionzky, S. Categories of socialbehavior. Comparative Group Studies, 1972, 3, 425-436.

Milgram, S. The experience of living in cities. Science,1970, 167, 1461-1468.

Nichols, K. A., & Champness, B. G. Eye gaze and theGSR. Journal of Experimental Social Psychology,1971, 7, 623-626.

Patterson, M. L. Compensation in nonverbal immediacybehaviors: A review. Sociometry, 1973, 36, 237-252.

Patterson, M. L. An arousal model of interpersonal in-^ timacy. Psychological Review, 1976, 83, 235-245.

Patterson, M. L. Arousal change and cognitive labeling:^ Pursuing the mediators of intimacy exchange. Envi-

ronmental Psychology and Nonverbal Behavior, 1978,3, 17-22. (a)

Patterson, M. L. The role of space in social interaction.In A. Siegman & S. Feldstein (Eds.), Nonverbal be-havior and communication. Hillsdale, N.J.: Erlbaum,1978 (b)

Patterson, M. L. Personality and nonverbal involvement:A functional analysis. In W. Ickes & E. S. Knowles(Eds.), Personality, roles, and social behavior. NewYork: Springer-Verlag, in press.

Patterson, M. L., Jordan, A., Hogan, M. B., & Frerker,D. Effects of nonverbal intimacy on arousal and be-havioral adjustment. Journal of Nonverbal Behavior,1981, 5, 184-198.

Patterson, M. L., Roth, C. P., & Schenk, C. Seatingarrangement, activity, and sex differences in smallgroup crowding. Personality and Social PsychologyBulletin, 1979, 5, 100-103.

Patterson, M. L., & Schaeffer, R. E. Effects of size andsex composition on interaction distance, participation,and satisfaction in small groups. Small Group Be-havior, 1977, 8, 433-442.

Rosenfeld, H. M. Conversational control functions ofnonverbal behavior. In A. W. Siegman & S. Feldstein(Eds.), Nonverbal behavior and communication.Hillsdale, N.J.: Erlbaum, 1978.

Rubin, A. Measurement of romantic love. Journal ofPersonality and Social Psychology, 1970, 16, 265-273.

Schachter, S., & Singer, J. E. Cognitive, social, and* physiological determinants of emotional state. Psy-

chological Review, 1962, 69, 379-399.

Page 19: A Sequential Functional Model of Nonverbal Exchangeipcommunication.wikispaces.com/file/view/non+verbal+exchange.pdf · A Sequential Functional Model of Nonverbal Exchange ... communication,

NONVERBAL EXCHANGE 249

Watzlawick, P., & Beavin, J. Some formal aspects ofcommunication. American Behavioral Scientist, 1967,10, 4-8.

Wegner, D. M., & Guiliano, T. Arousal-induced atten-tion to self. Journal of Personality and Social Psy-chology, 1980, 38, 119-726.

Whitcher, S. J., & Fisher, J. D. Multidimensional re-action to therapeutic touch in a hospital setting. Jour-nal of Personality and Social Psychology, 1979, 37,87-96.

Wicker, A. W. An introduction to ecological psychol-ogy. Monterey, Calif.: Brooks/Cole, 1979.

Wiener, M., Devoe, S., Rubinow, S., & Geller, J. Non-verbal behavior and nonverbal communication. Psy-chological Review, 1972, 79, 185-214.

Zajonc, R. B. Social facilitation. Science, 1965,16,269-274.

Zajonc, R. B. Feeling and thinking: Preferences needno inferences. American Psychologist, 1980, 35,151-175.

Received July 16, 1981Revision received January 8, 1982