Journal oj Personality and Social Psychology1968, Vol. 9, No. 3, 245-250

SOCIAL FACILITATION OF DOMINANT RESPONSESBY THE PRESENCE OF AN AUDIENCE AND

THE MERE PRESENCE OF OTHERS1

NICKOLAS B. COTTRELLUniversity of Iowa

GARY J. SEKERAKYale University

AND

DENNIS L. WACKPennsylvania State University

ROBERT H. RITTLEKent State University

Experiments have shown that the presence of an audience affects individual per-formance by enhancing the emission of dominant responses. An experimentwas conducted to evaluate the proposal of Zajonc that the mere presence ofother persons is responsible for audience effects. A total of 45 university stu-dents performed a pseudorecognition task; 15 performed the task alone, ISperformed the task before an audience of 2 passive spectators, and 15 per-formed the task in the presence of 2 persons who were not spectators. Thetask placed previously established verbal habits in competition with eachother. The presence of an audience enhanced the emission of dominant re-sponses, but the mere presence of others did not.

In a recent review, Zajonc (1965) usedHull-Spence theory to integrate the contra-dictory results from social facilitation studies.He proposed that the mere presence of otherpersons enhances the emission of dominantresponses by increasing the individual's gen-eral drive (D) level. Since dominant taskresponses may be either correct or incorrect,depending on the task and stage of practice,this proposal can accommodate both the socialincrements and the social decrements thathave been found in performance. Viewed inthis way, the effects of audience and coactionconditions upon individual performance aremerely instances of a process which occurswhenever other persons are present.

Recent studies of audience effects, usingpreviously validated behavioral indicators ofgeneral drive, have obtained results whichare consistent with the Zajonc proposal (Cot-trell, Rittle, & Wack, 1967; Zajonc & Sales,1966). These studies, however, do not show

1 This research was carried out at Kent State Uni-versity and was supported by Research Grants GS-1016 and GS-1956 from the National Science Founda-tion under the direction of Nickolas B. Cottrell.Some of these findings were presented by NickolasCottrell to the Miami University Symposium onSocial Behavior, April, 1967, Oxford, Ohio, and atthe September, 1967, meetings of the AmericanPsychological Association in Washington, D. C.

that the mere presence of others is responsi-ble for audience effects upon performance.

The aim of the present study was to deter-mine whether the presence of persons whoare not spectators or coactors also producesdrive effects upon individual performance.The present study compared the effects ofthree conditions—alone, mere presence, andaudience—upon performance on a pseudo-recognition task. Other persons were presentwhen the subject performed in both the audi-ence condition and the mere presence condi-tion, but only in the audience condition didthey have the status of spectators. In con-trast to the spectators in the present studyand in studies of audience effects (for in-stance, Cottrell et al., 1967; Zajonc & Sales,1966), the stimulus persons in the merepresence condition did not express interest inwatching the subject perform, and they wereunable to see the task stimuli to which thesubject responded.

The task used in this study places verbalhabits of different strengths in competitionwith each other. Zajonc and Sales (1966) usedthis task and found that the presence of anaudience enhanced the emission of responsesgoverned by strong habits at the expense ofresponses governed by weaker habits. Theproposal of Zajonc that the mere presence ofother persons increases drive level implies that

245

246 COTTRELL, WACK, SEKERAK, AND KITTLE

in the present study both the mere presencecondition and the audience condition shouldenhance the emission of dominant responses.

METHODSubjects

The subjects were 45 male introductory psychologystudents at Kent State University, participating tofulfill a course requirement. They were assignedrandomly to the three conditions of the experimentwith the restriction that IS participate in eachcondition.2

Apparatus and MaterialsThe stimuli were 10 nonsense words—AFWORBU,

BIWONJI, CIVADRA, JTEVKANI, LOKANTA, MECBURI, NAN-SOMA, PARITAF, SARiDiK, and ZABDLON—similar to thoseused by Zajonc and Sales (1966). The training stim-uli were 4 X 6-inch photos of each word. The teststimuli were 2 X 2-inch slides of each word.

The slides were presented on a Lafayette KT-800tachistoscope. A Meylan Electric Stopclock, 2022NF, was used to time the stages of the experiment.

ProcedureTraining. Verbal habits of varying strengths were

established by manipulating the number of times thesubject vocalized each of the nonsense words. Therewere five training frequencies: 1, 2, 5, 10, and 25.The 10 words were divided into five pairs. For eachsubject one of the word pairs was assigned to eachof the training frequencies. For each subject a deckof 86 photos was prepared consisting of 50 photosfor the two 25-frequency words, 20 photos for thetwo 10-frequency words, 10 photos for the two 5-frequency words, 4 photos for the two 2-frequencywords, and 2 photos for the two 1-frequency words.The cards were arranged in sequence by shuffling thedeck. The word pairs were rotated through thetraining frequencies in a 5 X 5 Latin-square whichwas completed three times in each experimental con-dition.

The experimenter described training as a studyof how people learn a foreign language. He pre-sented the photos by displaying a word, and thenreading it aloud. Then the subject read the wordaloud once. His pronunciation was neither reinforcednor corrected. The presentation rate was one photoevery 4 seconds.

Testing. During testing the verbal habits were2 Fourteen other students reported to the lab, but

were not used. One was blind, one was a member ofa racial minority, and one was excluded from theaudience condition because he discovered that theresponse words were not being presented on thepseudorccognition trials. Eleven students were ex-cluded—audience (4), mere presence (3), and alone(4)—because their responses on the first block ofpseudorecognition trials indicated they had notlearned more than three of the response words.

placed in competition with each other and the effectsof the experimental conditions upon the frequencyof emission of the verbal responses were observed.The subject received recognition instructions. Hewas also told that the speed of presentation wouldvary, that sometimes the word would be difficult toidentify, and on these trials he should guess. Toprevent rehearsal, the subject then read a passagefrom a history book.

The slides were tachistoscopically projectedthrough a small window in the control room upon ascreen in the adjacent experimental room which was7 feet from the subject. The two rooms were con-nected by an intercom. To familiarize himself withthe procedure, the subject first saw and named fourEnglish words—CLEVELAND, STUDENT, UNIVERSITY, andPROFESSOR—twice each at varying speeds and dia-phragm settings. Then the test stimuli were pre-sented in four blocks of 40 trials each. If a subjectfailed to respond within 10 seconds, the experimenterurged him to guess, and did not show the next slideuntil he did.

Each block consisted of 10 recognition trials and30 pseudorecognition trials. In each block the recog-nition trials consisted of single presentations of eachof the 10 training words. They were exposed with amedium diaphragm setting and a shutter speed of .2second. In pretests, these conditions produced cor-rect recognition on 85% of trials. The position ofthe recognition trials in the trial block and thetraining frequency of the word shown were deter-mined randomly. A different schedule was preparedfor each trial block and they were administered toall subjects.

On the pseudorecognition trials a stimulus waspresented which was an equally adequate stimulus forall of the 10 verbal responses. One of the trainingwords was exposed in reverse position with a smalldiaphragm opening for .01 second. In pretests, sub-jects reported seeing something wordlike under theseconditions, but were unable to identify words andcould not distinguish forward presentations frombackwards presentations. The subject's response wasaccepted if two of its syllables matched a trainingword. On the few occasions when the experimentercould not classify a response, he told the subjectthat his response was not one of the foreign wordshe had learned and asked him to make anotherguess.

The experimenter remained silent during the testtrials unless the subject was tardy in responding ormade an unclassifiable response.

Dependent variable. The dependent variable wasthe number of times words of each training fre-quency were emitted on the pseudorecognition trials.The pseudorecognition trials served to place theverbal habits established during training in competi-tion with one another. The subject was instructed tocall out 1 of the 10 training words on each trial andto guess when he was unsure. Since stimulation onthe pseudorecognition trials was equally adequatefor all of the 10 verbal responses, the guessing re-sponses on these trials showed the effective strength

SOCIAL FACILITATION OF DOMINANT RESPONSES 247

of the verbal habits when in competition with oneanother.

Experimental ConditionsIn the alone condition the subject was alone in

the experimental room during testing.In the audience condition two interested spectators

observed the subject and the test stimuli to whichhe was responding. Two confederates entered theexperimental room immediately after the subjectfinished reading the history passage. They posed asfellow introductory psychology students who werecoming to participate in a color-perception experi-ment. The experimenter informed them that theirexperiment would begin when the subject was fin-ished. The confederates obtained the experimenter'spermission to watch the present experiment whilethey waited. The experimenter instructed the subjectand the confederates not to talk to each other, andleft the room. The confederates sat 6 feet from thesubject and were positioned so they could observeboth the subject and the screen on which the wordswere projected. The confederates watched quietlyand attentively throughout the 160 test trials.

The mere presence condition differed from theaudience condition in that the two confederates didnot express interest in watching the subject, andthey wore blindfolds which prevented them fromobserving the stimuli to which the subject responded.After the subject had finished reading, the two con-federates entered, posing as subjects for a color-per-ception experiment. The experimenter noted thattheir experiment would begin when the subject wasfinished, and asked them to put on blindfolds to

14

19

12

a 1H(AZ 10

n 9-

O O ALONE• • MERE PRESENCE• » AUDIENCE

I 2 5 10 25TRAINING FREQUENCY

(HABIT STRENGTH)

FIG. 1. Number of responses of different training-frequency classes emitted on the pseudorecognitiontrials, averaged over subjects and over trial blocks.

eUIU)

§ '

tfum

UO<£5

TRIAL BLOCK I TRIAL BLOCK U

TRIAL BLOCK m 2-

TRAINING FREQUENCY (HABIT STRENGTH)• • AUDIENCE• • MERE PRESENCEO o ALONE

FIG. 2. Number of responses of different trainingfrequency classes emitted in separate blocks ofpseudorecognition trials, averaged over subjects.

adapt to the dark in preparation for their experi-ment. After he instructed the subject and confed-erates not to talk, the experimenter returned to thecontrol room. The blindfolded confederates wereseated in the same proximity to the subject as inthe audience condition and waited silently throughthe 160 test trials.

The confederates were male undergraduates. Theywere assigned to the conditions with the restrictionthat each confederate participate in both conditionsan approximately equal number of times.

To equalize the interval between training andtesting, subjects in the alone condition read morelines of the history passage than did subjects in theaudience and mere presence conditions.

RESULTSResponse Emission on PseudorecognitionTrials

Figure 1 presents for the three conditionsthe mean number of responses of each train-ing-frequency class emitted on the pseudo-recognition trials. There were IS subjects ineach condition and each contributed 120 re-sponses (4 blocks of 30 pseudorecognitiontrials). The response frequencies were ave-raged over trial blocks and subjects.

248 COTTRELL, WACK, SEKERAK, AND RITTLE

TABLE 1

SLOPE OP FUNCTION RELATING RESPONSE EMISSION TOTRAINING FREQUENCY FOR THREE CONDITIONS

SEPARATELY AND COMBINED ON SEPARATEAND COMBINED TRIAL BLOCKS

Trials

Block IBlock 11Block IIIBlock IVResponses

averagedover allblocks

3conditionscombined

.428

.359

.336

.257

.346

Alone

.363

.235

.226

.202

.257

Merc

.321

.328

.283

.235

.292

Audience

.600

.515

.500

.334

.488

Note.—There were IS subjects in eacli condition. Responsefrequencies were averaged over subjects.

Figure 2 presents the mean number of re-sponses of each training-frequency classemitted on each block of pseudorecognitiontrials. The data summarized in Figure 2 weresubmitted to a three-factor (testing condition,training frequency, and trial block) analysisof variance. Three effects were significant.The main effect of training frequency issignificant (p < .001, F = 44.30, df = 4/168). Analysis of the trend components3 re-vealed that the linear component of thetraining frequency effect is significant (p <.001, 7^ = 90.50, d /= l /42) , and accountsfor 88"ye of the variation between trainingfrequency classes. Response emission on thepseudorecognition trials is an increasing linearfunction of training frequency. The slope ofthis function is .346.

The purpose of the present study was toevaluate the drive-producing (D) effects ofthe presence of an audience and the merepresence of others. Hull-Spence theory(Spence, 1956; Spence & Spence, 1966) as-sumes that in a competitional situation theprobability of a response's emission is a directfunction of the magnitude of its excitatorypotential (E) relative to those of other re-sponses. The E variable is assumed to be amultiplicative function of the habit (H) anddrive (D) variables. Since in the present ex-periment the values on the habit factor—thetraining frequencies—were the same for all

8 Gaito's (1965) suggestions were followed inderiving trend coefficients for unequal intervals.

testing conditions, variation in the drive fac-tor would produce an interaction betweenconditions and training frequency by changingthe slope of the response-emission function.The analysis of variance showed that theinteraction between conditions and trainingfrequency is significant (p < .025, F = 2.38,rf/ = 8/168). This interaction is primarilydue to differences between conditions in lineartrend. The significant (p < .05, F = 3.90, df= 2/42) linear component of the interactionaccounted for 71 % of the interaction sums ofsquares. Comparison of the slopes of the re-sponse-emission functions shown in Figure 1by analysis of the linear trend showed thatthe response-emission function for the audi-ence condition is significantly steeper than itis for both the alone condition (F = 7.21, df= 1/28, ^ < .025) and the mere presencecondition (F = 4.53, df = 1/28, p < .05).The slope of the response-emission functionfor the mere presence condition was not re-liably different ( / ?< ! ) from the slope forthe alone condition.

The effects of the experimental conditionsupon response emission were similar over alltrial blocks, since the three-factor interactionof testing condition, training frequency, andtrial blocks was not significant (F = 1.21, df= 24/504, ns). Figure 2 shows that on everytrial block the presence of an audience en-hanced the emission of responses governed bystrong habits at the expense of responses gov-erned by weaker habits, and that on everytrial block response emission in the merepresence condition was similar to responseemission in the alone condition. Table 1 showsthat on every trial block the slope of the re-sponse-emission function in the audience con-dition was greater than the slope in the othertwo conditions, and also that on every trialblock the slope for the mere presence condi-tion was similar to the slope for the alonecondition.

Table 1 also shows that on successive trialblocks the slope of the response-emission func-tion for all conditions combined became lesssteep. This finding is reflected in the thirdsignificant effect of the analysis of variance, asignificant (p < .001, F ~ 3.00, df = 12/504)interaction between training frequency andtrial blocks. Significant (p < .001, F = 7.83,

SOCIAL FACILITATION OF DOMINANT RESPONSES 249

df = 3/216) variation between trial blocks inthe slope of the response-emission functionaccounted for 83% of the interaction sums ofsquares.

Accuracy on Recognition TrialsOn 40 trials—10 in each block—the train-

ing stimuli were presented supraliminally. Thethree conditions did not differ (F < 1) in themean percentage of correct recognitions (aloneX - n.2%, mere presence X = 79.8%, audi-ence X - 80.7%) on these trials.

Time MeasuresThe three conditions did not differ by t test

either in the mean number of minutes to com-plete the training stage of the experiment(alone X = 6.0, mere presence % — 6.1, audi-ence X — 6.1) or in the mean number of min-utes to complete the testing stage (alone X— 29.3, mere presence X = 28.6, audience X= 28.3). However, the mean interval betweentraining and testing was significantly shorterby t test (p < .005) in the alone condition(X — 6.2) than in both the mere presencecondition (% = 7.0) and the audience condi-tion (X - 6.9).* Thus it is possible that theshorter training-testing interval was responsi-ble for the testing performance of the alonesubjects, rather than their solitude duringtesting. Analyses of covariance were performedfor the comparisons involving the alone con-dition. The length of the training-testinginterval was the covariate, The covarianceadjustments were small—in all analyses thepooled within-group regression coefficient waszero to three decimals—and the adjustmentsdid not change the pattern of findings reportedabove. Therefore, the shorter training-testinginterval does not seem to be responsible forthe performance of the alone subjects.

DISCUSSIONThe results of the present study show that

the presence of an audience increases the in-4 Bringing the confederates into the testing room

in the audience and mere presence conditions re-quired more time in the experiment than in pretests.Thus, subjects in the alone condition did not spendsufficient time in reading from the history passageto equate the training-testing interval among condi-tions.

dividual's general drive (D) level. On thepseudorecognition trials the presence of anaudience enhanced the emission of dominantresponses at the expense of subordinate re-sponses. Previous research (Zajonc & Nieu-wenhuyse, 1964) has shown that this is theeffect of drive manipulations upon responseemission on the pseudorecognition trials.

The findings of the present study corre-spond quite closely to the findings of Zajoncand Sales (1966). This study also showed thatthe presence of an audience enhances theemission of dominant responses on thepseudorecognition trials. The slope of the re-sponse-emission function for subjects testedalone was .368 5 in the present study and .308in the Zajonc and Sales (1966) study. Theslope for subjects tested with an audiencewas .699 in the present study and .669 in theZajonc and Sales (1966) study.

Another similarity in the findings of thetwo studies is that in both studies there wasa significant interaction between training fre-quency and trial blocks upon response emis-sion due to a decline in the slope of the re-sponse-emission function over successive trialblocks. The decline results from the supralimi-nal exposure of each of the 10 training stim-uli once during every trial block, serving toincrease the usage of the low-frequency wordsas guesses.

The results of the present study do notsupport the proposal of Zajonc (1965) thatthe mere presence of other persons is responsi-ble for audience effects upon performance. Al-though the presence of an audience enhancedthe emission of dominant responses, the merepresence of other persons of the same status

5 The slopes of the present study shown in Table1 were multiplied by the factor 43/30 to make themcomparable to those obtained by Zajonc and Sales(1966). This procedure was necessary because in theZajonc and Sales (1966) study the measurementunit on the ordinate was the same width as themeasurement unit on the abscissa, but in the presentstudy the ordinate unit was wider than the ab-scissa unit. In the Zajonc and Sales (1966) studythere were 31 pseudorecognition trials per block andthe sum of the training frequency values was also31, but in the present study there were 30 pseudo-recognition trials per block and the sum of thetraining frequency values was 43.

250 COTTRELL, WACK, SEKERAK, AND RITTLE

(subjects for a later experiment) and in thesame physical proximity as the audience didnot enhance the emission of dominant re-sponses. The slope of the response-emissionfunction for the mere presence condition wasnot reliably different from the slope for thealone condition, and the response-emissionslope for the audience condition was signifi-cantly steeper than the slope for the merepresence condition.

Further research may show that the Za-jonc proposal holds true only for audienceand coaction arrangements; drive effects uponindividual performance will not occur unlessthe others present are either spectators or co-actors. Since simple physical presence doesnot seem to be sufficient, further researchmust also determine the factor which is re-sponsible for audience effects upon individualperformance.

REFERENCESCOTTRELL, N. B., RITTLE, R. H., & WACK, D. L.

Presence of an audience and list type (competi-tional or noncompetitional) as joint determinantsof performance in paired-associates learning.Journal of Personality, 1967, 35, 425-434.

GAITO, J. Unequal intervals and unequal n in trendanalysis. Psychological Bulletin, 1965, 63, 125-127.

SPENCE, K. W. Behavior theory and conditioning.New Haven: Yale University Press, 1956.

SPENCE, J. T., & SPENCE, K. W. The motivationalcomponents of manifest anxiety: Drive and drivestimuli. In C. D. Spielberger (Ed.), Anxiety andbehavior. New York: Academic Press, 1966.

ZAJONC, R. B. Social facilitation. Science, 1965, 149,269-274.

ZAJONC, R. B., & NIEUWENHUYSE, B. Relationshipbetween word frequency and recognition: Per-ceptual process or response bias? Journal of Ex-perimental Psychology, 1964, 67, 276-285.

ZAJONC, R. B., & SALES, S. M. Social facilitation ofdominant and subordinate responses. Journal ofExperimental Social Psychology, 1966, 2, 160-168.

(Received November 13, 1967)