Equilibrium Theory Revisited: Mutual Gaze and Personal Space in ...

Jeremy N BailensonJim BlascovichAndrew C BeallJack M LoomisDepartment of PsychologyUniversity of California

Santa Barbara CA 93106

Presence Vol 10 No 6 December 2001 583ndash598

copy 2001 by the Massachusetts Institute of Technology

Equilibrium Theory RevisitedMutual Gaze and Personal Spacein Virtual Environments

Abstract

During the last half of the twentieth century psychologists and anthropologists have

studied proxemics or spacing behavior among people in many contexts As we

enter the twenty-rst century immersive virtual environment technology promises

new experimental venues in which researchers can study proxemics Immersive

virtual environments provide realistic and compelling experimental settings without

sacricing experimental control The experiment reported here tested Argyle and

Deanrsquos (1965) equilibrium theoryrsquos specication of an inverse relationship between

mutual gaze a nonverbal cue signaling intimacy and interpersonal distance Partici-

pants were immersed in a three-dimensional virtual room in which a virtual human

representation (that is an embodied agent) stood Under the guise of a memory

task participants walked towards and around the agent Distance between the par-

ticipant and agent was tracked automatically via our immersive virtual environment

system All participants maintained more space around agents than they did around

similarly sized and shaped but nonhuman-like objects Female participants main-

tained more interpersonal distance between themselves and agents who engaged

them in eye contact (that is mutual gaze behavior) than between themselves and

agents who did not engage them in eye contact whereas male participants did not

Implications are discussed for the study of proxemics via immersive virtual environ-

ment technology as well as the design of virtual environments and virtual humans

1 Equilibrium Theory Revisited Mutual Gaze and PersonalSpace in Virtual Environments

Proxemics the study of personal space and interpersonal distance beganmore than four decades ago Hall (1959) and Sommer (1959) demonstratedthat people maintain personal or buffer space around themselves and eachother Although the size of the buffer space remains remarkably stable acrossindividuals certain conditionsmdashfor example nonverbal expressions of intimacy(such as mutual gaze)mdashfoster its expansion or contraction Argyle and Dean(1965) describe the interaction between mutual gaze and proxemic behaviorsAccording to their intimacy equilibrium model the two behaviors are inverselyrelated to each other Mutual gaze nonverbally promotes intimacy which ifinappropriate to the relationship between interactants is decreased by increasesin personal space (which nonverbally promotes less intimacy)

Immersive virtual environments (IVEs) raise at least two intriguing issues forproxemics research One involves the validity of using IVE technology (IVET)

Bailenson et al 583

to study proxemics experimentally If IVET is method-ologically valid for this purpose then it provides a pow-erful research tool Investigators can study proxemicswith complete control over virtual human representa-tions while at the same time maintaining a relativelyhigh degree of ecological validity and mundane realism(Aronson amp Carlsmith 1969) A second issue concernsIVEs as a new type of space that may itself affect non-verbal and proxemic behaviors within them

11 Nonverbal Communication

Patterson (1995) denes nonverbal communica-tion as the ldquotransmission of information and inuenceby an individualrsquos physical and behavioral cuesrdquo (p424) Nonverbal communication has been studied ex-tensively in psychology (for a review see Argyle(1988)) anthropology (Hall 1966 Watson 1970)and computer science (Badler Chi amp Chopra 1999Isbister amp Nass 2000) Nonverbal signals can be ex-pressed through many channels ranging from subtleones such as voice intonation to more obvious onesinvolving hand gestures Moreover these behaviors areoften subconscious and unintentional (Zajonc 1980)

12 Proxemics

Personal space the distance between two or morehuman beings has primarily been studied experimen-tally in one of four ways chair selection in which partic-ipants choose seats that vary in distance from a targetperson stop distance in which participants indicatewhen a real person such as an experimenter or confeder-ate should stop approaching them projective studies inwhich participants manipulate dolls and gures andnatural observational studies (See Hayduk (1983) for areview) Researchers have identied several other factorsthat moderate personal space including culture (Hall1966 Watson 1970) race (Rosegrant amp McCroskey1975) physiology (McBride King amp James 1965)age (Willis 1966) and interpersonal relationships(Evans amp Howard 1973 Little 1965)

Some researchers have argued that proxemic behav-iors differ for men and women Specically they claim

that personal space between men is the largest betweenwomen is the smallest and between men and women ismidlevel Several experiments have demonstrated thatcompared to men women maintain less space betweenthemselves and other people (Adler amp Iverson 1974Aiello 1977) have bodies that take up smaller amountsof physical space (Mehrabian 1972 Jenni amp Jenni1976) and are more likely than men to withdraw whentheir space is invaded (Henley 1977) However theevidence for these effects is mixed In a survey of theproxemics literature Hayduk (1983) determined that27 studies found sex differences in the size of personalspace 54 studies found mixed evidence and 29 studiesfound no effects

IVEs may offer an experimental media that can helpinvestigators examine gender differences in proxemicbehaviors more reliably Virtual environments offer aunique space for examining interactions because experi-menters can maximize realism as in a eld study withoutsacricing the experimental control of a laboratory(Loomis Blascovich amp Beall 1999) Many proxemicsstudies have relied on naturalistic observation (whereexperimental control is at a minimum) confederates(whose behavior is necessarily variable) or projectivetechniques (which are largely unrealistic) to answerquestions concerning personal space Perhaps questionsof personal space can be investigated more scienticallyusing IVEs Research measuring the proxemic behaviorof icons on two-dimensional desktop environments(Krikorian Lee Chock amp Harms 2000) suggests thatpeople do attempt to monitor their personal space invirtual environments However whether they do reli-ably in immersive three-dimensional virtual environ-ments with realistic humanoid representations remainsto be seen

13 Mutual Gaze

Mutual gaze occurs when two people are lookingat each otherrsquos eyes Linguistically mutual gaze helpspeople organize interactions by regulating conversa-tional sequencing (Argyle 1988) However mutualgaze transmits information above and beyond linguisticregulation Research demonstrates that people who ex-

584 PRESENCE VOLUME 10 NUMBER 6

hibit high levels of mutual gaze are perceived as inti-mate (Scherer amp Schiff 1973) attentive (Breed Chris-tiansen amp Larson 1972) competent (SodikoffFirestone amp Kaplan 1974) and powerful (Argyle Le-febvre amp Cook 1974) In addition people can be in-uenced by mutual gaze without necessarily beingaware of it (Zajonc 1980) Women tend to exhibitmore mutual gaze in dyadic interactions than do men(Argyle amp Cook 1976 Chapman 1975) In additionwomen tend to tolerate and more favorably react togaze than do men (Valentine amp Ehrlichman 1979)Furthermore in a recent study on pedestriansrsquo gazeavoidance Patterson amp Webb (in press) demonstratedthat men gaze more often at women than at other menbut that women tend to gaze at men and other womenequally as often Similarly numerous studies show thatwomen are more adept than men are at transmittingand receiving nonverbal information (See Hall (1984)for a review)

According to Argyle and Deanrsquos (1965) research onthe equilibrium theory mutual gaze (a nonverbal cuesignaling intimacy) moderates interpersonal distanceSeveral studies support this hypothesis RosenfeldBreck Smith and Kehoe (1984) explored various con-ditions in which confederates violated participantsrsquo per-sonal space by recording the number of times that theparticipants gazed at the confederatesrsquo eyes These re-searchers demonstrated a dramatic reduction in gaze inresponse to invasion of personal space Similarly Patter-son (1976 1982) and Hayduk (1981) demonstratedthat participants increase personal space between them-selves and confederates who increased mutual gazeMoreover participants will move closer when facing theconfederatesrsquo backs than their fronts (Aono 1981 Ash-ton amp Shaw 1980 Hayduk 1981)

Although there has been much research on mutualgaze little of that research has involved controlled rep-resentations of people Just as with proxemics IVEs canimprove the methods used to study mutual gaze Gazebehavior of virtual humans can be regulated to be lesssusceptible to error than the scripted gaze behavior ofconfederates Furthermore using IVEs we can guaran-tee that participantsrsquo eye height is exactly the same asthe eye height of the virtual human thus improving the

probability that the manipulations of gaze will be no-ticed (as well as eliminating potential status differencesdue to height)

Virtual humans when rendered stereoscopically inthree-dimensional virtual environments can prove to besurprisingly compelling representations of living hu-mans Recent conceptual and technological break-throughs allow us to create virtual human representa-tions that are behaviorally realistic (Badler et al 1999Massaro 1998 Cassel amp Thorisson 1998) In real hu-mans different nonverbal behaviors are often highlycorrelated with each other (Argyle 1988 Dittmann ampLlewellyn 1969) Hence it is likely that past studiesthat claimed to manipulate one behavior actually manip-ulated many For example it would be difcult for aconfederate to maintain eye gaze with a participantwithout moving his or her hands or slightly changingfacial expressions or breathing patterns Controllingthese behaviors in virtual humans allows us to maintaincomplete independence among these behaviors

Previously we dened the representation of someentity in a virtual environment as a virtual human Wecan distinguish an embodied agent from an avatar1 Anembodied agent is a virtual representation that is con-trolled entirely by a computer program whereas an ava-tar is a virtual representation that is controlled at leastpartially by a human being In the current study weexplore behavior with participants who are immersedwith only agents

2 The Experiment

21 Overview

We immersed experimental participants in a virtualroom in which a virtual male agent stood We instructedthem to remember certain features and labels on thefront and back of the agentrsquos shirt Unbeknownst to theparticipants we recorded their absolute position andorientation with a precision tracking system as theywalked about the virtual room

1 This distinction is equivalent respectively to ldquoagent-avatarrdquo andldquohuman-avatarrdquo the distinction raised by Blascovich et al (in press)

Bailenson et al 585

22 Hypotheses

We drew our hypotheses from Argyle and Deanrsquos(1965) equilibrium model By varying the degree ofmutual gaze between the agent and the participant wewere able to test the inverse relationship of mutual gazewith personal space We hypothesized that when theagent constantly maintained eye gaze with the partici-pant the participant would leave the agent a larger bub-ble of personal space compared to when the agent hadhis eyes closed In addition we varied the photographicrealism of the agentrsquos face the agent either had a tex-ture-mapped and photographically realistic face or onecreated from a series of atly shaded polygons Nowak(2000) also explores a similar difference in photographicrealism and calls it the degree of anthropomorphism Inline with Blascovich et alrsquos (in press) arguments regard-ing the relative importance of behavioral realism such asmutual gaze over photographic realism we predictedlarger effects on proxemic behavior due to mutual gazethan to photographic realism

Furthermore we predicted that the size and shape ofthe footprint of the personal space bubble maintainedaround agents would be similar to that maintainedaround real humans We hypothesized that participantswould maintain more space in front of the agent thanthey would behind it We did not make predictions con-cerning gender differences in distance behavior as theliterature is unclear in this regard (Hayduk 1983)However because the literature on mutual gaze moreclearly indicates that women notice mutual gaze morethan men do we hypothesized that women would re-spond more strongly to our manipulations of gaze be-havior than would men

23 Methods

231 Design We manipulated one within-partic-ipant variable (model of the agentrsquos face) and two be-tween-participant variables (gaze behavior of the agentand gender of the participant) There were two levels offace at shaded and photograph textured Flat-shadedfaces were ldquochiseledrdquo that is constructed as a three-dimensional model that had noticeably sharp facial to-

pography Photograph-textured faces had an image ofan actual human face tted to the three-dimensionalmodel loosely borrowing a technique developed bySannier and Thalmann (1998) Figure 1 gives examplesof the two conditions

We manipulated ve levels of increasingly realisticgaze behavior In the lowest level 1 the agentrsquos eyeswere closed In level 2 his eyes were open In level 3the agentrsquos eyes were open and he blinked In level 4 inaddition to blinking his eyes the agent turned his headso that he constantly gazed at the participantrsquos face as he

Figure 1 Pictures of the two face conditions The at shaded-face

is on top the photographic texture is on the bottom


or she traversed the environment The agentrsquos headturned 85 deg in either direction Level 5 of gaze be-havior was the same as level 4 however the agentrsquos pu-pils dilated by 50 when the participant stepped within075 m of the agent Figure 2 illustrates the differencesamong these levels In levels 4 and 5 the agent doesnot necessarily demonstrate completely realistic gazebehavior because he does not ever glance away fromthe participant We realize that normal gaze behaviorincludes random glances away from the target but tomaintain experimental control we did not include spo-radic movements such as glancing away Along the samelines we chose to separate these gaze behaviors in a sys-tematic way as opposed to integrating them all at onceIn this fashion we could attempt to gauge the uniquecontribution of each behavior

Each participant ldquointeractedrdquo with agents in a sin-gle level of gaze behavior We presented participantstwo blocks of trials one block with at polygon-shaded faces and one with photograph-textured facesEach block had ve trials and the order of blocks was

counterbalanced across participants In addition weran a control condition on a separate group of partici-pants that was identical to the other conditions ex-cept that instead of a humanoid representation in theroom there was an object representation (a pylon)that was the same width and height as the agent Thetask in this control condition was exactly the same asthe other conditions

232 Materials and Apparatus The virtualroom was modeled to be 72 m by 64 m by 45 m high(approximately 75 of the space of the physical room)to ensure that participants did not walk into any physicalwalls while they traversed the virtual room Figure 3shows the location of the agent in the virtual room aswell as the participantsrsquo starting point The agent wasrepresented as a Caucasian male three-dimensionalpolygon-based model His height was 185 m and hisbody was always facing south in the room He wore alabel on both the front and back of his shirt The frontlabel listed his name and the back label listed a num-

Figure 2 Pictures of the different gaze-behavior levels

Figure 3 An example of the ten paths from a participant in

realism condition 4 The ticks on the axes represent meters

Bailenson et al 587

ber2 The size of the text on each label was chosen sothat at a viewing distance of 1 m the task of readingthe word or number was perceptually easy with only afew quick xations

Participants themselves were not rendered Hencealthough participants could walk about the virtual envi-ronment and see the agent or the pylon in the roomthey did not see any animated representation of them-selves Consequently if a participant looked down whilewalking he or she would not see his or her own legsand feet We set the eye height to be exactly the same asthe agentrsquos eyes for all participants and participants be-gan the trial facing the agent (north in the room)

The technology used to render the IVEs is shown ingure 4 The head-mounted displays (HMD) was a Vir-tual Research V8 HMD (a stereoscopic display withdual 680 3 480 resolution LCD panels that refresh at72 Hz) The optics of this display presented a visualstimulus subtending approximately 50 deg horizontallyby 38 deg vertically Perspectively correct stereoscopicimages were rendered by a 450 MHz Pentium III dual-processor computer with an Evans amp Sutherland Tor-nado 3000 dual-pipe graphics card and these imageswere updated at an average frame rate of 36 Hz Thesimulated viewpoint was continually updated by the par-

ticipantrsquos head movements The orientation of the par-ticipantrsquos head was tracked by a three-axis orientationsensing system (Intersense IS300 update rate of 150Hz) and the location of the participantrsquos head wastracked three-dimensionally by a passive optical positionsensing system (developed in our lab and capable ofmeasuring position with a resolution of 1 part in30000 or approximately 02 mm in a 5 m squareworkspace 60 Hz) The system latency or the amountof delay between a participantrsquos head or body motionand the resulting concomitant update in the HMDrsquosvisual display was a maximum of 65 ms Using this hy-brid tracking system it is possible for a participant toexperience appropriate sensory input when she turns herhead at the same time as she walks There was no colli-sion detection In other words a participant could walkthrough the agent or through a virtual wall without re-ceiving any haptic or auditory cues

233 Participants Participants were recruitedon campus and were either paid or given experimentalcredit in an introductory psychology class for participa-tion Four men and four women participated in each ofthe ve gaze-behavior conditions and six men and fourwomen participated in the control condition resultingin fty total participants in the study Participantsrsquo ageranged from 18 to 31

234 Procedure One individual participated ineach session We instructed participants that they wouldbe walking around a room and engaging in a memorytest They read the following paragraph

In the following experiment you will be walkingaround in a series of virtual rooms In the rooms withyou will see a person The person is wearing a whitepatch on the front of his shirt His name is written onthat patch He is also wearing a similar patch on theback of his shirt On the back patch a number is writ-ten Your job is to walk over to the person in theroom and to read the name and number on hispatches First read the back patch and then read thefront patch Later on we will be asking you questionsabout the names and numbers of the person in each

2 The numbers were all prime and the names were all six letterstwo syllables and matched for frequency using Yahoo Web-basedwhite-pages software The names and numbers appear in appendix AThe text of the names and numbers were modeled to be approxi-mately 15 in high

Figure 4 A depiction of our virtual environment system The

components are (1) position tracking cameras (2) HMD and

orientation tracking sensor and (3) image generator


room We will also be asking you about their cloth-ing hair color and eye color When you have readthe patches and examined the person in each roomwe will ask you to step back to the starting point inthe room The starting point is marked by a piece ofwood on the oor

Our ostensible experimental task of reading andmemorizing the agentrsquos name and number motivatedthe participant to move within a relatively close range (1m or less) of the agent so as to easily read the textualmaterial We felt that by design this secondary taskwould unwittingly cause the subject to move closeenough to the avatar as to intrude potentially upon thehypothesized personal space bubble of this entity Sub-sequently the participantrsquos movements would resultfrom a competition between their desire to maintain anappropriate level of personal space and their need toaccurately read the patches

When they understood the instructions participantstried on the HMD To become accustomed to theequipment and walking while immersed participantswere given a chance to walk around an empty virtualroom while wearing the HMD Participants freely ex-plored the room for approximately 1 min and thenwalked back to the starting point None of the partici-pants had any trouble nding the starting point

After the practice exploration participants began therst block Figure 5 simulates a participant walkingaround the virtual room For each trial participants be-gan facing the agent They then stepped from the start-ing point and walked around to the back of the agent(or the pylon in the control condition) They read thenumber on the agentrsquos back and then walked backaround to his front After reading the front patch theyreturned back to the starting point and waited for thenext trial We chose the patch-reading task for two rea-sons First it ensured that our participants would bemotivated to approach the agent relatively closely Sec-ond because nonverbal behavior tends to be implicitand often unintentional (Zajonc 1980) we decidedthat measuring personal space while subjects were inten-tionally engaged in a distracting task would be the mosteffective manner to elicit these behaviors

For each of the ve trials in a block the agent wore adifferent-colored shirt had different-colored hair andhad a different name and number In the control condi-tion the pylon was colored differently for each trial (thesame colors as the agentrsquos shirts in the other condi-tions) Across participants the names numbers andother features appeared in each serial trial position anequal number of instances Blocks took between 5 and15 min depending on the participantrsquos walking rateParticipants had an opportunity to rest between blocks

After participants completed the two blocks they re-moved the HMD and were administered a pen-and-paper recall test For the recall test participants tried toldquorecall all the names and numbers on the patchesrdquo Af-ter the recall test participants received a matching testin which all the names and numbers were listed Their

Figure 5 The virtual environment as seen by the user In this gure

we include a representation of the user wearing the HMD solely to

indicate scale However participants could not see a representation of

their own avatar

Bailenson et al 589

task was to draw lines that connected the name of theagent on a specic trial to the number of that agent onthe same trial We instructed participants to draw all tenlines guessing when they were unsure if a name wentwith a number

Finally after the recall test participants put the HMDback on for two more trials in order to complete a socialpresence questionnaire one with the photograph-tex-tured face and one with the at-shaded face We neverexplicitly instructed our participants that the avatar wasan agent controlled entirely by the computer Howeverpostexperimental interviews indicated that none of ourparticipants suspected the avatar was controlled by an-other human being

For the survey a Likert-type scale (from 2 3 to 1 3)hung in space over the agentrsquos head Participants lookedat the agent and the scale while the experimenter ver-bally administered the ve-item social presence ques-tionnaire People feel high social presence if they are ina virtual environment and behave as if interacting withother veritable human beings For a more detailed dis-cussion of social presence in immersive virtual environ-ments see Blascovich et al (in press) To capture themost realistic measure of social presence possible weasked the ratings questions while participants were im-mersed The questions appear in appendix B Partici-pants in the control condition did not answer the ques-tionnaire

24 Results

Participants had no problems walking through thevirtual space and none experienced any signicant sim-ulator sickness After the experiment none of the partic-ipants indicated that they had guessed that their prox-emic behaviors were under scrutiny All were under theimpression that we were primarily studying memory

The tracking system saved the participantrsquos position ata rate of 18 Hz Figure 3 depicts the paths that a typicalparticipant traversed over the ten trials Each positionsample located the participantrsquos position in the virtualroom For each participant we recorded the minimumdistance between the centerpoint of the participantrsquoshead and the centerpoint of the agentrsquos head during

each trial There were no reliable differences betweenthe two types of faces (photograph textured and atshaded) in any of the analyses Consequently we col-lapsed across this variable in subsequent analyses

241 Personal Space Our primary predictionsconcerned the distance between the participants and theagent Figures 6 and 7 show the paths by participantsrsquogender and experimental condition Participants actuallystepped through the agent on only two trials (by twodifferent participants) out of 400 Interestingly both ofthem were in the lowest level of gaze behavior (that isthe agentrsquos eyes were closed) The other 38 participantsdid not ldquotouchrdquo the agent

We used two objective measures of participantsrsquo inva-sion of the agentrsquos personal space minimum distanceand invasion duration Minimum distance was denedas the shortest distance that participants maintained be-tween themselves and the agent We chose minimumdistance instead of average distance for two reasonsrst as Hayduk (1983) points out many previous stud-ies measuring proxemics relied on this measure andsecond because some participants concentrated onreading the labels they spent blocks of time at a specicreading distance Consequently given the nature of thetask average distance may not accurately reect partici-pantsrsquo attention to the nonverbal gaze behavior On theother hand minimum distance is a better measure ofhow close they were willing to go to the agent whileexamining his features and walking around him Thesecond measure invasion duration was dened as theamount of time that participants spent inside the agentrsquosintimate space which based on work by Hall (1966)we dene as the number of seconds spent within a rangeof 45 cm The invasion duration data were almost iden-tical to the minimum distance data such that the twomeasures produced similar signicant effects and alsocorrelated highly with each other Consequently for thepurposes of brevity and clarity we present only the min-imum-distance data It is important to note here thatour distance measure may result in larger distances thanother measures (see Hayduk (1981)) because we mea-sure distance between the centerpoint of the heads in-stead of the perimeter of the heads


Regarding personal space and gaze behavior we pre-dicted monotonic increases with increasing realism Weexpected that participants would be most likely to re-spect an agentrsquos personal space when he exhibited realis-tic gaze behaviors (Argyle amp Dean 1965) Conse-quently we predicted that minimum distance should belongest in the high-gaze conditions (4 and 5) and short-est in the condition in which the agentrsquos eyes wereclosed and in the control condition in which there is acylindrical pylon instead of a humanoid agent (See hy-potheses)

We analyzed both back minimum distance (the mini-mum distance while the participant was behind the mid-point of the agentrsquos head) and front minimum distance(the minimum distance while the participant was infront of the midpoint of the agentrsquos head) The averageback minimum distance (not including the control con-dition) was 037 m (sd 5 015) the minimum was

006 and the maximum was 068 The average frontminimum distance was 040 (sd 5 015) the mini-mum was 004 and the maximum was 071 A one-wayANOVA indicated a marginal effect for the differencebetween front and back minimum distances F(1 39) 5

325 p 008 which is consistent with studies dis-cussed previously (Hayduk 1983) that nd the foot-print of the personal space bubble to be slightly largerin front than in back The two measures correlated witheach other signicantly r 5 078

Table 1 displays the average back minimum distanceby condition Participants approached more closely tothe back of the cylinder (M 5 024 sd 5 006) thanto the back of the agent (M 5 037 sd 5 013) F (148) 5 1027 p 001

Table 1 suggests that female participants did in factexhibit the predicted monotonic trend To further ex-plore that trend we ran a nonparametric correlation

Figure 6 A plot of the female participantsrsquo paths by condition The graphs include only data that

surrounds the agent Each point on the gure represents a sample reading (taken at 20 Hz) Each

gure depicts the samples from all of the subjects from that condition

Bailenson et al 591

test3 between participantsrsquo back minimum distance andtheir gaze-behavior level First we examined female par-ticipants With 24 observations (four female participantsin each of the ve levels plus the control) we nd a sig-nicant Spearmanrsquos correlation r 5 047 p 001 Wealso nd a signicant effect for male participants r 5

043 p 005 As predicted both men and womengave more personal space to agents who exhibited real-istic gaze behavior than to agents who did not as well asto the control pylon

We then used a regression procedure to partial outthe variance due to memory matching scores and socialpresence ratings (see subsection 242) from back mini-mum distance Cohen and Cohen (1983) provide a for- mal description of this process The purpose of this

analysis was twofold to statistically control for partici-pants whose high motivation in the memory task pre-vented them from noticing the agentrsquos gaze behaviorand to statistically control for participants who found

3 We used a nonparametric correlation for this analysis because thelevels of the realism variable were ordinal For this reason we couldnot run a simultaneous analysis using presence matching and realism(as a linear variable)

Table 1 Average and Standard Deviation of Back Minimum

in Meters by Condition

Gaze Condition Women Men

Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)

Figure 7 A plot of the male participantsrsquo paths by condition The graphs include only data that




the entire virtual reality experience to be either too in-timidating or too unpersuasive to notice the agentrsquosgaze behavior In later sections we discuss both mem-ory scores and social presence ratings individually

Table 2 shows the standardized means (adjusted formemory matching and social presence) of minimumback distance by condition We again ran Spearmanrsquoscorrelation test this time between the participantsrsquo stan-dardized adjusted back distance mean and their gaze-behavior level First we examined female participantsWith 24 observations4 we still nd signicant effects forboth women (r 5 047 p 001) and men (r 5 036p 005)

Table 3 shows the means of front minimum distanceby condition Participants approached the front of thecylinder more closely (M 5 031 sd 5 009) than thefront of the agent (M 5 040 sd 5 015) F (1 48) 5

374 p 0059 We ran the same Spearmanrsquos correla-tion analysis between front minimum distance and gaze-behavior level that we did for back minimum distanceThere were no signicant effects for either female partic-ipants (r 5 033) or for male participants (r 5 022)However again we adjusted the means for social pres-ence and memory matching score Table 4 shows theadjusted standardized means of minimum front distanceby condition The correlation between adjusted frontdistance and gaze-behavior level was signicant for

women (r 5 34 p 0055) but not for men (r 5

2 013) When taking into account social presence andmemory task motivation women gave more personalspace to the front of agents who exhibited realistic gazebehavior

Interestingly when controlling for social presenceand matching we found a marginal effect for socialpresence ratings across male and female subjectst(19) 5 182 p 009 Further examination demon-strated a Pearsonrsquos correlation of 044 p 0054 be-tween social presence ratings and front minimum dis-tance for men5 In other words male participants didnot adjust their degree of front minimum distance ac-

4 We did not have presence ratings for the control condition sowe partialed out only the variance from matching score for those ob-servations

5 Using back minimum distance the correlation between socialpresence ratings and minimum distance is in the same direction formen (r 5 038) but this effect was not signicant with only twentyobservations

Table 4 Average Standardized Means of Covariate-AdjustedFront Minimums by Condition


Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015

Table 2 Average Standardized Means of Covariate-AdjustedBack Minimums by Condition


Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044

Table 3 Average and Standard Deviation of Front Minimumin Meters by Condition


Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593

cording to our gaze-behavior manipulations Insteadthey respected the space of agents to which they subjec-tively assigned sentience and gaze in their social pres-ence ratings This effect is consistent with data fromPatterson and Webb (in press) that demonstrates thatmen do not maintain eye contact with other men andconsequently would not notice male gaze as much aswomen would This evidence explains why table 2 doesnot show a clear linear trend for male participants

Another interesting nding was the degree of individ-ual differences by mutual gaze condition As gures 6and 7 demonstrate more variance occurred in the lowgaze than in the high-gaze behavior conditions To testthis hypothesis we implemented Levenersquos test for ho-mogeneity of variances using gaze as an independentvariable For the dependent variable we included thedistance between every point illustrated in gures 6 and7 and the agent With all six levels of gaze behaviorthere was no signicant effect However when welumped the two conditions (four and ve) in which theagent maintained constant mutual gaze we see thatthere was signicantly less intersubject variance in thehigh-gaze behavior conditions (073 cm) than in thelow-gaze behavior conditions (183 cm) F(1 48) 5

450 p 005 In other words participants displayedlarge individual differences showing high variance inregards to how far away they stand from the agent whenthe agent did not maintain eye contact with themHowever their proxemic behavior was more uniform(that is less variant) in the high-gaze conditions

In sum women are more affected by the gaze behav-iors of the agent and adjust their personal space moreaccordingly than do men However men do subjec-tively assign gaze behavior to the agent and their prox-emic behavior reects this perception Furthermoreboth men and women demonstrate less variance in theirproxemic behavior when the agent displays mutual gazebehavior than when the agent does not

242 Social Presence Ratings On the socialpresence ratings task we summed the ten responses intoa simple score Cronbachrsquos reliability alpha a measurecommonly used to assess reliability across the individualitems of a scale (Cohen amp Cohen 1983) was 083

across the ten questions A positive social presence scoreindicates that the participant believed the agent wasconscious and was watching him or her whereas a nega-tive score indicates that the participant felt the agentwas just a computerized image The average social pres-ence rating score was 2 518 (sd 5 1142) the mini-mum was 2 30 and the maximum was 16 Fifteen out offorty scores were positive indicating that more than athird of the participants perceived the agent to be realis-tic and assigned some degree of sentience to him

Just as with personal distance there was absolutely noeffect of the face variable or gender Table 5 shows theaverage rating score by condition We ran a nonpara-metric correlation between gaze behavior and ratingscore Across gender Spearmanrsquos rho was 030 p

003 indicating that participants experienced highersocial presence when the agentrsquos gaze behavior washigh We then ran the same analysis by gender The cor-relation was not signicant for the 24 men (r 5 024)but it was for the 24 women (r 5 042 p 003) Thefact that women tailored their social presence ratings tothe gaze manipulations is consistent with studies (Hall1984) that demonstrate that women are more likely toaccurately decode nonverbal gestures than men are

243 Memory Participantsrsquo recall tended to bepoor in general so for the remainder of the section wediscuss their memory score on the matching task (per-centage of correct matches) Neither gaze behavior norgender signicantly affected matching score Howeverwhen we compared the control condition to the otherve levels of gaze behavior we found that participantsrsquomemory was better for the control condition with the

Table 5 Average Social Presence Score by Condition


1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)


pylon in the room (M 5 028 sd 5 18) than for thetrials with the agent in the room (M 5 015 sd 5

011) F(1 48) 5 724 p 001 This result is consis-tent with previous work on social inhibition (BlascovichMendes Hunter amp Salomon 1999) in which the pres-ence of another person impeded performance on novelor difcult tasks6

25 Discussion

The focus of this research is on the subtle nonver-bal exchanges that occur between a person and an em-bodied agent Participants in our study clearly did nottreat our agent as a mere animation On the contrarythe results suggest that in virtual environments peoplewere inuenced by the three-dimensional model Ini-tially participants overwhelmingly avoided direct con-tact with the agent despite the fact that our system didnot employ any type of display to indicate collisionsMoreover participants respected personal space of thehumanoid representation more than they did the cylin-der in the control condition

Furthermore we found gender differences in responseto nonverbal gestures that are consistent with ndingsfrom the nonvirtual environment literature Female partici-pants noted the gaze behavior of the agent more thanmale participants did as indicated by the social presenceratings Furthermore not only did women respond moreto the nonverbal behaviors in their ratings but they werealso more likely to respect the agentrsquos personal space if hedisplayed realistic gaze behavior than were the males in ourstudy This conrmation of the equilibrium theory in vir-tual environments is particularly notable considering thatthe participants had absolutely no idea that we were mea-suring their personal space and that the gaze behavior con-dition was varied between subjects Furthermore theseresults occurred in a social situation that was completelynonverbal in other words our participants treated theagent in a manner similar to the manner in which they

would treat humans despite the fact that there were noactual verbal exchanges

In addition to modulating our participantsrsquo proxemicbehaviors the agent affected their performance on thememory task Participants who encoded the names andnumbers while the humanoid agent was in the roomhad difculty on the memory task (compared to thecontrol condition) This social inhibition on memorylends additional support to the notion that the presenceof virtual observers affected people in similar ways toreal observers

3 Implications and Future Research

Immersive virtual environments are becomingmore commonplace in the elds of social psychologicalresearch (Slater Sadagic Usoh amp Schroeder 2000Loomis et al 1999 Hoyt 1999 Swinth 2000) com-munication (Guye-Vuilleme Capin Pandzic Thal-mann amp Thalmann 1999 Biocca amp Levy 1995) andbusiness (DeFanti 2000) It seems inevitable that as weuse these virtual environments more and more interac-tions between avatars will become routine Our data issimilar to previous work (Reeves amp Nass 1996) thatshows that people do assign some degree of sentience toa relatively simple virtual representation of a human be-ing especially when that representation exhibits realisticgaze behavior Furthermore as Blascovich et al (inpress) argue realistic gaze behavior seems to be morecrucial than photographic realism in establishing thesocial presence of an agent This result has implicationsfor the design of agents as well as for virtual environ-ments in general

The current study provides a starting point for an em-pirical exploration of nonverbal gestures of agents andavatars in IVEs We have established that proxemics canbe a valuable tool for measuring the behavioral realismof an agent or an avatar People tend to perceive non-verbal gestures on an implicit level (Zajonc 1980) anddegree of personal space appears to be an accurate wayto measure peoplersquos perception of social presence andrealism in virtual environments Other studies demon-strate that implicit behavioral measures such as body

6 We do acknowledge the possibility that this effect occurred be-cause there was an extra feature to remember on the human gure(hair color) than on the pylon and that this extra feature may havesomehow distracted participants

Bailenson et al 595

posture can be a reliable measure of the userrsquos sense ofpresence in virtual environments (Freeman AvonsMeddis Pearson amp IJsselsteijn 2000) Similarly per-sonal space may be a more reliable measure of socialpresence than a typical ratings survey in immersive vir-tual environments

As discussed in the previous paragraphs our data rep-licate a number of ndings from the eld of nonverbalgestures and personal space (Argyle 1988) Conse-quently our study suggests validation for the use of vir-tual environments as a medium to study human behav-ior Moreover virtual environment technology appearsto deliver the exceptional balance of realism and experi-mental control it promised Consequently nonverbalbehaviors that previously could be studied using onlyeld observations (such as gaze and personal space) cannow be rigorously explored in the laboratory

One shortcoming of the current study is that theagent whom participants approached was always maleEspecially because we found an effect of the partici-pantrsquos gender it is crucial for future studies to exam-ine peoplersquos spacing behavior towards both male andfemale agents Along similar lines future studiesshould address the distinction between agents andavatars in virtual environments In the current studywe employed agents that were driven by the com-puter However participantsrsquo use of personal space

might be vastly different when they approach an ava-tar that is directly controlled by a human being Weare currently testing this hypothesis

Appendix B The questions from the socialpresence survey

1 I perceive that I am in the presence of anotherperson in the room with me

2 I feel that the person is watching me and is awareof my presence

3 The thought that the person is not a real personcrosses my mind often

4 The person appears to be sentient (conscious andalive) to me

5 I perceive the person as being only a computerizedimage not as a real person

Acknowledgments

The authors would like to thank David Waller and Max Weis-buch for comments on this paper as well as Meg BrzezinskaKelly Eversole Trisha Gordon Michael Gunanto Mike Rai-mundo and Mike Yao for assistance in collecting data andediting Furthermore we thank Kim Swinth and Crystal Hoytfor help in developing materials This research was sponsoredin part by NSF Award SBE-9873432

References

Adler L L amp Iverson M A (1974) Interpersonal distanceas a function of task difculty praise status orientation andsex of partner Perceptual amp Motor Skills 39 683ndash 692

Aiello J R (1977) A further look at equilibrium theory Vi-sual interaction as a function of interpersonal distance Envi-ronmental Psychology amp Nonverbal Behavior 1 122ndash140

Argyle M (1988) Bodily Communication (2nd ed) LondonMethuen

Argyle M amp Cook M (1976) Gaze and Mutual Gaze NewYork Cambridge University Press

Argyle M amp Dean J (1965) Eye-contact distance and af-liation Sociometry 28 289 ndash304

Argyle M Lefebvre L amp Cook M (1974) The meaning

Appendix A The Names and Numbers from the Agentsrsquo

Labels

Names Numbers

Stader 37Borken 29Waller 19Jotter 5Mervis 73Nerson 61Veltis 23Rudins 11Debnar 3Kipple 53


of ve patterns of gaze European Journal of Social Psychol-ogy 4 125ndash136

Aronson E amp Carlsmith J M (1969) Experimentation insocial psychology In G Lindzey amp E Aronson (Eds)Handbook of social psychology (2nd ed vol II) (pp 1ndash79)Reading MA Addison-Wesley

Badler N I Chi D amp Chopra S (1999) Virtual humananimation based on movement observations and cognitivebehavior models Proceedings of SIGGRAPHrsquo99 1ndash17

Biocca F amp Levy M (1995) Communication in the Age ofVirtual Reality Lawrence Erlbaum Associates Inc Hills-dale NJ

Blascovich J Loomis J Beall A Swinth K Hoyt C ampBailenson J N (In press) Immersive virtual environmenttechnology as a methodological tool for social psychologyPsychological Inquiry

Blascovich J Mendes W B Hunter S B amp Salomon K(1999) Social ldquofacilitationrdquo as challenge and threat Journalof Personality amp Social Psychology 77 68ndash77

Cassel J amp Thorisson K R (1998) The power of a nodand a glance Envelope vs emotional feedback in animatedconversational agents Journal of Applied Articial Intelli-gence 13 519 ndash538

Chapman A (1975) Eye contact physical proximity andlaughter A re-examination of the equilibrium model of so-cial intimacy Social Behavior amp Personality 3 143ndash155

Cohen J amp Cohen R (1983) Applied Multiple RegressionCorrelation Analysis for the Behavioral Sciences HillsdaleNJ Lawrence Erlbaum Associates Inc

DeFanti T (2000) Better than being there Next millenniumnetworks IEEE 20 60 ndash63

Dittmann A T amp Llewellyn L G (1969) Body movementand speech rhythm in social conversation Journal of Person-ality amp Social Psychology 11 98ndash106

Evans G W amp Howard R B (1973) Personal space Psy-chological Bulletin 80 334ndash344

Freeman J Avons S E Meddis R Pearson D E amp IJs-selsteijn W (2000) Using behavioral realism to estimatepresence A study of the utility of postural responses to mo-tion stimuli Presence Teleoperators and Virtual Environ-ments 9 149 ndash164

Guye-Vuillieme A Capin T K Pandzic I S ThalmannN M amp Thalmann D (1999) Non-verbal communica-tion interface for collaborative virtual environments TheVirtual Reality Journal 4 49 ndash59

Hall E T (1959) The Silent Language NY Doubledaymdashmdashmdash (1966) The Hidden Dimension NY Doubleday

Hall J A (1984) Nonverbal sex differences Accuracy of com-munication and expressive style Baltimore John HopkinsUniversity Press

Hayduk L A (1981) The shape of personal space An exper-imental investigation Canadian Journal of Behavioural Sci-ence 13 87ndash93

mdashmdashmdash (1983) Personal space Where we now stand Psycho-logical Bulletin 94 293ndash335

Henley N M (1977) Body politics Power sex and nonverbalcommunication Englewood Cliffs NJ Prentice-Hall

Hoyt C (1999) Social facilitation in virtual environments Anew methodological tool for social psychologists Unpublishedmasters thesis [Instr] University of California Santa Bar-bara

Isbister K amp Nass C (2000) Consistency of personality ininteractive characters Verbal cues non-verbal cues anduser characteristics International Journal of Human-Com-puter Studies 53 251ndash267

Jenni D A amp Jenni M A (1976) Carrying behavior inhumans Analysis of sex differences Science 194 859 ndash860

Krikorian D H Lee J Chock T M amp Harms C (2000)Isnrsquot that spatial Distance and communication in a 2-Dvirtual environment Journal of Computer-Mediated Com-munication [online] 4 wwwascuscortjcmcvol5issue4krikorianhtm

Little K B (1965) Personal space Journal of ExperimentalSocial Psychology 1 237ndash247

Loomis J Blascovich J amp Beall A (1999) Immersive vir-tual environment technology as a basic research tool in psy-chology Behavior Research Methods Instruments amp Com-puters 31 557ndash564

Massaro D W (1998) Perceiving Talking Faces From speechperception to a behavioral principle Cambridge MA MITPress

McBride G King M G amp James J W (1965) Socialproximity effects on galvanic skin responses in adult hu-mans Journal of Psychology 61 153ndash157

Mehrabian A (1972) Nonverbal communication ChicagoAldine-Atherton

Nowak K (2000) The Inuence of Anthropomorphism onMental Models of Agents and Avatars in Social Virtual Envi-ronments Unpublished doctoral dissertation Departmentof Telecommunication Michigan State University

Patterson M L (1976) An arousal model of interpersonalintimacy Psychological Review 83 235ndash245

mdashmdashmdash (1982) A sequential functional model of nonverbalexchange Psychological Review 89 231ndash249

Bailenson et al 597

mdashmdashmdash (1995) Invited article A parallel process model ofnonverbal communication Journal of Nonverbal Behavior19 3ndash29

Patterson M L amp Webb A (in press) Passing encountersPatterns of recognition and avoidance in pedestrians Basicand Applied Social Psychology

Reeves B amp Nass C (1996) The media equation How peo-ple treat computers television and new media like real peopleand places New York Cambridge University Press

Rosegrant T S amp McCroskey J C (1975) The effects ofrace and sex on proxemic behavior in an interview settingThe Southern Speech Communication Journal 40 408ndash 420

Rosenfeld H M Breck B E Smith S H amp Kehoe S(1984) Intimacy-mediators of the proximity-gaze compen-sation effect Movement conversational role acquaintanceand gender Journal of Nonverbal Behavior 8 235ndash249

Sannier G amp Thalmann M N (1998) A user friendly tex-ture-tting methodology for virtual humans Proceedings ofthe Computer Graphics International (CGI rsquo97)

Scherer S E amp Schiff M R (1973) Perceived intimacyphysical distance and eye contact Perceptual amp Motor Skills36 835ndash 841

Slater M Sadagic A Usoh M amp Schroeder R (2000)Small-group behavior in a virtual and real environmentPresence Teleoperators and Virtual Environments 9 37ndash51

Sodikoff C L Firestone I J amp Kaplan K J (1974) Dis-tance matching and distance equilibrium in the interviewdyad Personality amp Social Psychology Bulletin 1 243ndash245

Sommer R (1959) On writing ldquolittle papersrdquo American Psy-chologist 14 235ndash237

Stafford L Kline S L amp Dimmick J (1999) Home e-mail Relational maintenance and gratication opportuni-ties Journal of Broadcasting amp Electronic Media 43 659 ndash669

Swinth K (2000) Risk-taking and conformity in immersivevirtual environments Unpublished masters thesis Univer-sity of California Santa Barbara

Valentine M E amp Ehrlichman H (1979) Interpersonalgaze and helping behavior Journal of Social Psychology 107193ndash198

Waller D A (in press) Individual differences in spatial learn-ing in computer simulated environments Journal of Experi-mental Psychology Applied

Watson O M (1970) Proxemic behavior A cross-culturalstudy The Hague Mouton

Willis F N Jr (1966) Initial speaking distance as a functionof the speakersrsquo relationship Psychonomic Science 5 221ndash222

Zajonc R B (1980) Feeling and thinking Preferences needno inferences American Psychologist 35 151ndash175


to study proxemics experimentally If IVET is method-ologically valid for this purpose then it provides a pow-erful research tool Investigators can study proxemicswith complete control over virtual human representa-tions while at the same time maintaining a relativelyhigh degree of ecological validity and mundane realism(Aronson amp Carlsmith 1969) A second issue concernsIVEs as a new type of space that may itself affect non-verbal and proxemic behaviors within them

11 Nonverbal Communication

Patterson (1995) denes nonverbal communica-tion as the ldquotransmission of information and inuenceby an individualrsquos physical and behavioral cuesrdquo (p424) Nonverbal communication has been studied ex-tensively in psychology (for a review see Argyle(1988)) anthropology (Hall 1966 Watson 1970)and computer science (Badler Chi amp Chopra 1999Isbister amp Nass 2000) Nonverbal signals can be ex-pressed through many channels ranging from subtleones such as voice intonation to more obvious onesinvolving hand gestures Moreover these behaviors areoften subconscious and unintentional (Zajonc 1980)

12 Proxemics

Personal space the distance between two or morehuman beings has primarily been studied experimen-tally in one of four ways chair selection in which partic-ipants choose seats that vary in distance from a targetperson stop distance in which participants indicatewhen a real person such as an experimenter or confeder-ate should stop approaching them projective studies inwhich participants manipulate dolls and gures andnatural observational studies (See Hayduk (1983) for areview) Researchers have identied several other factorsthat moderate personal space including culture (Hall1966 Watson 1970) race (Rosegrant amp McCroskey1975) physiology (McBride King amp James 1965)age (Willis 1966) and interpersonal relationships(Evans amp Howard 1973 Little 1965)

Some researchers have argued that proxemic behav-iors differ for men and women Specically they claim

that personal space between men is the largest betweenwomen is the smallest and between men and women ismidlevel Several experiments have demonstrated thatcompared to men women maintain less space betweenthemselves and other people (Adler amp Iverson 1974Aiello 1977) have bodies that take up smaller amountsof physical space (Mehrabian 1972 Jenni amp Jenni1976) and are more likely than men to withdraw whentheir space is invaded (Henley 1977) However theevidence for these effects is mixed In a survey of theproxemics literature Hayduk (1983) determined that27 studies found sex differences in the size of personalspace 54 studies found mixed evidence and 29 studiesfound no effects

IVEs may offer an experimental media that can helpinvestigators examine gender differences in proxemicbehaviors more reliably Virtual environments offer aunique space for examining interactions because experi-menters can maximize realism as in a eld study withoutsacricing the experimental control of a laboratory(Loomis Blascovich amp Beall 1999) Many proxemicsstudies have relied on naturalistic observation (whereexperimental control is at a minimum) confederates(whose behavior is necessarily variable) or projectivetechniques (which are largely unrealistic) to answerquestions concerning personal space Perhaps questionsof personal space can be investigated more scienticallyusing IVEs Research measuring the proxemic behaviorof icons on two-dimensional desktop environments(Krikorian Lee Chock amp Harms 2000) suggests thatpeople do attempt to monitor their personal space invirtual environments However whether they do reli-ably in immersive three-dimensional virtual environ-ments with realistic humanoid representations remainsto be seen

13 Mutual Gaze

Mutual gaze occurs when two people are lookingat each otherrsquos eyes Linguistically mutual gaze helpspeople organize interactions by regulating conversa-tional sequencing (Argyle 1988) However mutualgaze transmits information above and beyond linguisticregulation Research demonstrates that people who ex-








2 The Experiment

21 Overview



Bailenson et al 585

22 Hypotheses



23 Methods














Bailenson et al 587






















their own avatar

Bailenson et al 589




24 Results
















Bailenson et al 591









Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597

























2 The Experiment

21 Overview



Bailenson et al 585

22 Hypotheses



23 Methods














Bailenson et al 587






















their own avatar

Bailenson et al 589




24 Results
















Bailenson et al 591









Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597



















22 Hypotheses



23 Methods














Bailenson et al 587






















their own avatar

Bailenson et al 589




24 Results
















Bailenson et al 591









Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597


























Bailenson et al 587






















their own avatar

Bailenson et al 589




24 Results
















Bailenson et al 591









Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597








































their own avatar

Bailenson et al 589




24 Results
















Bailenson et al 591









Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597






















24 Results
















Bailenson et al 591









Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597




























Bailenson et al 591









Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597



























Control 025 (004) 022 (005)1 032 (026) 036 (018)2 033 (006) 042 (016)3 036 (001) 027 (006)4 042 (005) 042 (008)5 045 (015) 036 (010)
















Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597






























Control 2 031 2 1211 2 062 2 0362 2 038 0363 2 008 2 0744 052 2 0485 046 2 015



Control 2 078 2 1851 2 006 0032 2 014 0643 002 2 0654 058 0525 059 2 044



Control 037 (006) 027 (012)1 034 (029) 037 (016)2 039 (006) 048 (018)3 039 (008) 030 (006)4 050 (012) 035 (015)5 046 (013) 041 (011)

Bailenson et al 593












1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597






























1 2 1325 (1858) 2 1375 (714)2 2 175 (547) 2 700 (1233)3 2 300 (1644) 2 525 (956)4 2 275 (454) 100 (939)5 25 (1555) 2 85 (645)




25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597





















25 Discussion









Bailenson et al 595











Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597





























Acknowledgments


References








Labels

Names Numbers


































Bailenson et al 597


















































Bailenson et al 597



















Equilibrium Theory Revisited: Mutual Gaze and Personal Space in ...

Documents

Transcript of Equilibrium Theory Revisited: Mutual Gaze and Personal Space in ...