Avatars in Social Media

 

ARTICLE ARTIC LE IN PR PRESS ESS

Int. J. Human-Computer Studies 66 (2008) 801–811 www.elsevier.com/locate/ijhcs

Avatars in social media: Balancing accuracy, playfulness and embodied messages Asimina Vasaloua,, Adam Joinsona, Tanja Ba ¨ ¨ nzigerb, Peter Goldiec, Jeremy Pittd a

School of Information Management, University of Bath, BA2 7AA, Bath, UK  Department of Psychology, University of Geneva, 40 Boulevard du Pont-d’Arve, CH-1205, Geneva, Switzerland  c Department of Philosophy, University of Manchester, Dover Street Building, M13 9PL, Manchester, UK  d Electrical and Electronic Engineering, Imperial College, Exhibition Road, SW7 2AZ, London, UK 

b

Received 9 May 2008; received in revised form 6 August 2008; accepted 11 August 2008 Communicated by S. Wiedenbeck Available online 14 August 2008

Abstract

This paper examines how users negotiat negotiatee their selfself-present presentation ation via an avata avatarr used in social media. Twenty partici participants pants customised customised an avatar while thinking aloud. An analysis of this verbal data revealed three motivating factors that drive self-presentation: (1) avatars were use used d to accura accuratel tely y refl reflect ect their their owne owners’ rs’ offl offline ine self; self; par partic ticipan ipants ts chose chose to disp display lay stable stable sel self-a f-attr ttribut ibutes es or idea idealise lised d their their ava avatar tar by concealing or emphasising attributes aligned to imagined social roles, (2) the diversity of customisation options was exploited by some participants partic ipants who broke free from the social rules governing self-pre self-presentat sentation ion offline; others used the avata avatar’s r’s appearance to emotio emotionally nally provoke and engage the avatar viewer and finally, (3) avatars were used as proxies; participants designed their online self in order to convey a message to a significant other. r 2008 Elsevier Ltd. All rights reserved. Self-presentation; Identity Keywords:  Avatars; Social media; Self-presentation;

1. Introd Introduction uction

Increasing Increa singly, ly, soc social ial sys system tem des design igners ers are int introd roduci ucing ng avatars as the new forum for identity construction in online enviro env ironmen nments, ts, a fea featur turee users users hav havee welcome welcomed d with enthusiasm. thusia sm. Following Yahoo’s launch of Yahoo! Avatars Avatars,, 7 million users personalised an avatar for display within their profi profile less e. e.g. g. Ya Yaho hoo! o! 36 360 01, Yaho Yahoo! o! An Answ swer ers, s, Ya Yaho hoo! o! Messen Mes senger ger (Hem Hemp, p, 200 2006 6). Like Likewis wise, e, four four milli million on us user erss created an avatar after the facility (called ‘weemee’) was introduced introd uced to the chat system ‘Windows Live Messenger’.1 This rapid uptake of avatars is not surprising; while avatars maintain users’ privacy, they give expressive freedom over 

Corresponding author. a.vasalou@bath [email protected] .ac.uk (A. Vasalou) Vasalou),,   a.joinson@ E-mail addresses: addresses:   a.vasal bath.ac bat h.ac.uk .uk (A. Join Joinson) son),,   Tanja.Banzig [email protected] [email protected] nige.ch (T. Ba ¨ ¨ nziger) nziger),, [email protected] (P. Goldie), Goldie),   [email protected] (J. Pitt). Pitt). 1Retrieved ¼

from http://www.weeworld.com/about/partner.aspx?partner 22 on 31/03/08.

1071-5819/$ 1071-58 19/$ - see front matter matter r 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhcs.2008.08.002 doi:10.1016/j.ijhcs.2008.08.002

an ot othe herw rwise ise st stat atic ic onlin onlinee id iden entit tity. y. In an il illu lustr strat ativ ivee example exa mple,, V-Chat V-Chat,, a vir virtual tual chat-r chat-room oom,, allowed allowed users users to either design their own avatar, which was particularly time demand dem anding, ing, or to select select from from alr alread eady-m y-made ade opt option ions. s. In genera gen eral, l, V-Chat V-Chat users users spurne spurned d the lat latter ter option option,, ins instead tead taking tak ing the time to create create uni unique que repres represent entatio ations ns so that that they ‘‘didn’t look like others’’ (Cheng (Cheng et al., 2002). 2002 ). Avat Av atar arss are are malle malleab able le ob obje jects cts as us user erss ar aree give given n th thee abilit abi lity y to choose choose diverse diverse custom customisa isation tion combin combinati ations ons.. Motivated by this fluid property of avatars, HCI researchers have have exa examine mined d the social social implica implicatio tions ns of avatars avatars by considering consi dering how users manage self-presenta self-presentation tion through through th this is new me medi dium um and and to what what exte extent nt an avata vatar’ r’ss appearance can shape online communication (e.g.  Bers et al., 2001; 2001;  Axelsson, 2002 2002;;  Taylor, 2002; 2002;  Nowak and Rauh, 2005). 2005 ). This prior work has tradition traditionall ally y taken taken place place in virtual environments where users’ avatars serve as online embodi emb odiment mentss for naviga navigating ting threethree-dime dimensi nsiona onall spaces spaces.. Though members may forge friendships over time, at the onset they communicate with previously unknown others.

 

ARTIC TICLE LE IN PR PRESS ESS AR 802

A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

More recently, avatars have been implanted in social media environments. Unlike the strangers one encounters in other online environments, users of social media predominantly interact with offline contacts such as friends and family: bloggers document their daily lives to keep their family and friends informed (Nardi (Nardi et al., 2004); 2004 ); users of the social networking site Facebook visit the site to reconnect with old friend friendss or to mainta maintain in ongoin ongoing g friend friendshi ships ps (Joinson,

self-presentation and (3) embodied messages. The theoretical implications of these results are extended to the field of online communication, while several practical implications are considered considered for the design of avatar avatar customisatio customisation n tools. 2. Backg Background round

2008). 2008 Furt Further hermor more, e, es thesel ope open nflectio tex text t nchanne cha nnel l ava availa ilable blecent int soc social ial). media med ia provok pro vokes self-re f-reflec tion that tha t is rem reminis iniscen of diary entries: bloggers document their life events (Nardi (Nardi et al al., ., 20 2004 04), ), us usee blog bloggin ging g as a form form of ‘‘ca ‘‘cath thar arsis sis’’ ’’ fo forr working out their emotions (Nardi (Nardi et al., 2004; 2004;   Huffaker and Calvert, 2005), 2005), or voice their opinions on controversial topics e.g. politics (Nardi (Nardi et al., 2004). 2004). Therefore, in social media, one’s offline social and emotional life is merged with the online sphere. Th This is unit unity y of offli offline ne an and d onli online ne has has been been foun found d to facilit fac ilitate ate mor moree honest honest con constr structi uctions ons of ide identit ntity y and self self-presen pre sentat tation ion,, as oppose opposed d to the role role pla playin ying g tha thatt oft often en ta take kess plac placee in vi virt rtua uall envi enviro ronm nmen ents ts or MU MUDs Ds (e (e.g .g.. Bruckman, Bruckman, 1993 1993;;   Turkl Turkle, e, 1995 1995;;   Taylo Taylor, r, 2002 2002). ).2 To giv givee an exa example mple,, blo blogge ggers rs freque frequently ntly report report ide identi ntifyi fying ng and

 2.1. Identity and self-presentation

accurate profile information such as their real name or age (Herr Herrin ing g et al al., ., 20 2004 04;;   Hu Huff ffak aker er an and d Ca Calv lver ert, t, 20 2005 05). ). A similar similar resul resultt has been sho shown wn in relatio relation n to ava avatars tars;; users tend to reflect their own appearance when personalising an avatar for display within their blog (Vasalou ( Vasalou et al., 2007). 2007 ). This This pre previo vious us wor work, k, how howeve ever, r, doe doess not revea reveall the st stra rate tegi gies es that that user userss draw draw toge togeth ther er to sh shap apee a self self-representative avatar. The present research addresses this point with a qualitative study that captures the array of  self-pr self -prese esenta ntatio tion n str strate ategies gies that that use users rs app apply ly whe when n conconstruc structi ting ng an avat avatar. ar. By eluc elucida idatin ting g the the ‘‘how ‘‘how’’ ’’ be behi hind nd self-pr self -prese esenta ntatio tion, n, we gain gain a mor moree pro profou found nd theore theoretica ticall understanding about identity in social media. At the same time,, by exp time exposi osing ng whi which ch custom customisa isatio tion n cho choice icess are sigsignificant to users, we extrapolate design recommendations for avatar customisation tools. We begin by discussing how theories on identity and selfpresen pre sentat tation ion apply apply to the unique unique qua qualiti lities es of the avatar avatar medium. med ium. Next, Next, we describ describee the experim experiment ent conduc conducted ted;; users were requested to construct an avatar for two social media scenarios. The qualitative ‘‘think aloud’’ methodology used in this work is also motivated. This is followed by an analysis of the results outlining the full spectrum of selfpresen pre sentat tation ion stra strateg tegies ies par partic ticipa ipants nts used used during during avatar avatar creatio cre ation. n. This This pap paper er ends ends with a gen genera erall dis discuss cussion ion that that identifies three motivating factors guiding self-presentation in social media: (1) accurate self-presentation (2) playful

stable characteristics such as age, gender and race. Other aspects of appearance are malleable such as one’s displayed social standing or recreational activities at a given moment in time. Manner is the expressive signals one intentionally or un unin inten tentio tiona nally lly gives gives of off. f. For For in inst stanc ance, e, a sp spea eake kerr presenting at a seminar, as expected, will display a calm and confident demeanour. demeanour. Conversely, Conversely, despite efforts to remain calm, the speaker may give off signals of anxiety. Generally, appearance and manner are expected to align. For example, during a project pitch, the creative director is formal for mally ly dressed dressed while while his his/her /her manner manner should should tra transmi nsmitt confidence with the team’s proposal even though the client may momentarily question his/her judgment. Unlike Goffman’s microscopic analysis of self,   Higgins (1987)   approached the same concept from a macroscopic view.. Accord view According ing to Higgin Higgins, s, there there are three three sel self-a f-aspe spects, cts, perceived either from the actor’s point of view or from an imagined audience’s perspective. The  ideal  self  self is composed of qualities which the actor or others wish for him/her to possess, while the  ought  self comprises qualities which the actor or others expect him/her to possess due to obligation. The  actual  self   self is composed of stable qualities the actor or others believe s/he possesses. Goffman’s and Higgins’ frameworks have been used to answ answer er ques questi tion onss re rega gard rdin ing g onlin onlinee id iden enti tity ty and and se self lf-on et al al.. presentati prese ntation. on. Aligned Aligned to Goffman’s Goffman’s view,   Ellis Ellison (2006) showed (2006)  showed that online daters have a tendency to idealise their self-presentation by controlling the cues they give off. In the wider field of computercomputer-media mediated ted communicat communication, ion, Bargh et al. (2002)   found found that online online users, users, liberated liberated by anonymity, generally express more actual self-aspects than when they communicate face-to-face. Identity construction in this previous work was limited to either text-based or pictorial picto rial representa representation tions. s. Avatars Avatars present present two distinctive distinctive

2

The qualitative differences between social media and virtual environments are also understood within  Schroeder’s (2002)   sociology-inspired framewor fram ework k for virt virtual ual environm environments ents:: an onli online ne interac interaction tion is uniq uniquely uely framed within a context and it is characterised by the social roles users are compe com pelle lled d to ta take. ke. In additi addition, on, th thee techn technolo ologic gical al featu features res of the the environment can encourage or discourage certain behaviours by influencing users’ object of focus.

Goffman (19 Goffman (1959) 59)   cons consid ider ered ed huma humans ns to be ac acto tors rs,, co cons nsta tant ntly ly fo follo llowin wing g a so socia ciall sc scrip ript. t. By us usin ing g   front, compos com posed ed of   setting,   appearance   and   manner, an acto actorr idea idealis lises es hi his/h s/her er se selflf-pr pres esen enta tatio tion n as s/ s/he he conc concea eals ls or emphasises aspects that correspond to the values governing a particular situation. More specifically, Goffman defined front fro nt as the, the, ‘‘expre ‘‘expressi ssive ve equipm equipment ent of a sta standa ndard rd kind kind intenti inte ntiona onally lly or unwitti unwittingl ngly y emp employ loyed ed by the individ individual ual Goffm fman an,, 19 1959 59,, p. 32 32). ). Th Thee during dur ing his perform performanc ance’’ e’’ (Gof setting is the stage which the actor chooses as a backdrop for his/her performance. A first date is arranged to take pl plac acee at a re rest stau aura rant nt;; a work work meet meetin ing g co conv nven enes es at a conference room. The actor’s appearance is comprised by

 

ARTIC LE IN PR PRESS ESS ARTICLE A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

properties, the opportunity for controlled expressivity and the the pres presen ence ce of (and (and poss possib ibil ility ity for) for) am ambi bigu guit ity. y. The The remainder of this section elaborates on how these properties may further influence users’ self-presentation.

 2.2. Controlled expressivity In fac face-to e-to-fa -face ce commun communicat ication ion,, the actor actor exe execut cutes es the pe perf rform orman ance ce coordinating in part part by appearance usin using g hi his/ s/he her r body bo dy as a tool tool,, intentionally and emotion expression (Goffman, (Goffman, 1959 1959). ). Though expression may constitute an intentional ‘‘signal’’ to others, it is also experienced as a 2001 1). ‘‘sy ‘‘ symp mpto tom’’ m’’ in th thee body body that that gene genera rates tes it (Hess, Hess, 200 Expressing Expres sing happiness at a wedding wedding reflects reflects the appropriate social script, while it simultaneously provokes arousal in one’s body. At the same time, expression is not under the actor’s act or’s complete complete contro controll as it oft often en fol follows lows a sub subject jective ive emotion experience. After a social blunder, the experience of em emba barr rras assm smen entt is spon sponta tane neou ousl sly y (a (ass oppo oppose sed d to strategically) expressed by blushing or by gaze avoidance 1997;;  Tracy and Robins, 2004). 2004). (Keltner and Buswell, 1997 Avat Av atar arss are are a pict pictori orial al repre represen senta tatio tion n of the the bo body dy.. Sy Symp mpto toms ms and and symb symbols ols are are no nott in inte terlo rlock cked ed into into one one biol biolog ogica icall syst system em.. Becau Because se of this, this, an avat avatar’ ar’ss signa signals ls might be created by its owner to align to his/her intended expres exp ression sion without without sponta spontaneo neously usly act activa ivatin ting g the correcorresponding emotion. When chatting online, one may laugh at another’s anoth er’s spelling error privately, privately, but may easily choose no nott to disp displa lay y the the em emot otion ion of am amus usem ement ent vi via a his/ his/her her avatar. This possibility also opens up more opportunities for misrepr misrepresen esentin ting g onesel oneself, f, as an unt untalen alented ted act actor or no longer needs to rely on his/her poor expressive talent, but on the in-built expressivity of the avatar. An avatar may reflect a happy demeanour despite its owner’s disgruntled personality. perso nality. Moreover, Moreover, online, online, there is an indefinite time window between the act of identity construction and the actual display of front which offers the user time to reflect on the precise qualities he/she wants to emit (Ellison ( Ellison et al., 2006). 2006 ). These qualities put together can contribute to selfpresentation that is exceptionally strategic and tailored to convey conve y precise messages.

 2.3. Ambiguity Avatar customisation environments are populated with options of fantasy, e.g. science-fiction costumes, as well as conventional and familiar real-life objects, e.g. jeans and a t-shirt. Often, these contrasting items are equally accessible and prominent in the interface. By contrast, in the physical world, an actor selects a setting for the performance that already exists in the real world, while clothes fit for the occasio occ asion n are worn. Rarely Rarely is the actor confro confronted nted with fictitio fict itious us items items for constru constructi cting ng his his/her /her front such such as a magi ma gicc wa wand nd or an al alie ien n ma mask sk.. Ther Theref efor ore, e, un unli like ke the the physical world, avatars introduce an element of ambiguity. Ambiguity and the space it opens up for interpretation has become an important design goal in HCI, especially for

 

803

social systems whose purpose is to provoke interpersonal dialogue (Gaver (Gaver et al., 2003; 2003;  Boehner and Hancock, 2006 2006). ). Regard Reg arding ing avatar avatars, s, the contin continuum uum of real real and fantas fantasy y invites users to reflect on their social roles, reinventing their social preconceptions. For example, an avatar may be used to challenge its owner’s traditional family role. This selfreflection can transcend into real life, shaping his/her future encoun enc ounter terss with with family family member memberss (Turkl Turkle, e, 1995 1995;;   Taylor, 2002 2002). ). Furthermore, user can designer using the avatar as a tool tothe provoke thebecome viewer;ainconsistencies and metaphors built into the avatar may create a narrative that challenges the viewer’s interpretation. 3. Metho Methods ds

The pre presen sentt res researc earch h str strive ived d to underst understand and the entire entire breadth and diversity of users’ self-presentation strategies when whe n creati creating ng an avatar avatar for social social media. Thus, it was important to have direct access to users’ thoughts during the process of customisation. In choosing a methodology, we dr drew ew fr from om th thee fie field ld of us usab abili ility ty whic which h ha hass re rece cent ntly ly focuse foc used d on creatin creating g novel novel method methodss for user user experie experience nce de desig sign n and and eval evalua uatio tion n (e (e.g .g.. se selflf-as asse sessm ssmen entt mani manikin kin:: 1994; experience clips: Isomursu Bradley and Lang, 1994; clips:  Isomursu et al., 2007;; int 2007 interv erviews iews::   Su Sund ndst stro rom m et al. al.,, 20 2007 07;; fu fuzz zzy y lo logi gicc emotion modelling: Mandryk modelling:  Mandryk and Atkins, 2007). 2007). The methods we came across did not allow us to collect qualitative data concurrently to users’ interaction with the avatar ava tar tool; tool; hence, hence, we borrowe borrowed d a method method tra traditi ditiona onally lly us used ed in us usab abil ility ity st stud udies ies,, th thee conc concur urre rent nt th thin ink k alou aloud d protoc pro tocol. ol. The thi think nk aloud aloud protoc protocol ol was suitab suitable le for our purposes as it requires users to verbalise their actions and thoughts while using an interface.

3.1. Parti Participan cipants ts The study was advertised in bulletin boards throughout the Univer University sity of Geneva Geneva.. In tot total, al, 20 partici participan pants ts took took part in the think aloud study, nine of whom were male. Partici Par ticipan pants ts were were aged aged betwee between n 20 and 40 yea years rs old, old, all weree experi wer experienc enced ed comput computer er partic participa ipants nts and non nonee had previo pre vious us experie experience nce with avatar avatars. s. As the ses session sionss were were conduc con ducted ted in French, French, only only fluent fluent French French speake speakers rs were recruited.

3.2. Proced Procedure ure An adapted version of the think aloud protocol, tailored for for the the need needss of us usab abil ilit ity y st stud udie ies, s, was was used used to el elic icit it participants partic ipants’’ verbalisation verbalisationss (see  (see   Boren and Ramey, 2000). 2000 ). When participants first arrived they were given a training task instructing them to think aloud by speaking out their onscre ons creen en action actions. s. The facilit facilitato atorr made made sure sure to positio position n hi hims msel elff as th thee li list sten ener er wh who o will will part partak akee li litt ttle le in th thee protocol, while the participant was given the role of the expert exp ert speake speaker. r. During During the ses session sion,, the facilit facilitato atorr reafreaffirmed firm ed his role role as the lis listen tener er by occasi occasiona onally lly provid providing ing

 

ARTIC TICLE LE IN PR PRESS ESS AR 804

A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

acknowledgment tokens (e.g. hmm). If the participant was silent for too long, the facilitator prompted him/her. Two observers located in a room adjacent to the lab viewed and recorded record ed the session. session. The recordings recordings were later used for the qualitative analysis. The avatar tool used in this study was Yahoo! Avatars (http://avatars.yahoo.fr). http://avatars.yahoo.fr). Yahoo! Avatars gives users control over their avatar’s body with variable facial features, eye colo colour, ur, In hairst haiaddition rstyle yle and colour, r, skin ski n ton tone, e, clothe clo thess and accessories. to acolou neutral emotional expression, an avata vatar’ r’ss fa face ce ca can n be assi assign gned ed to fo four ur di diff ffer eren entt expressions. expres sions. Furthermore, Furthermore, avata avatars rs can be placed against differe diff erent nt bac backgr kgroun ound d locatio locations. ns. Oth Other er art artefa efacts cts are also offere off ered d for the purpos purposee of person personalis alisati ation, on, such such as pet pets, s, flags, furniture items, among others. The options offered in Yahoo! Avatars cover a wide spectrum of everyday items, such as pets, hobbies and city locations, as well as fantasy options, such as masks and costumes. Therefore, choosing Yahoo! Avatars ensured that participants’ choices were not constrained by the possibly limited options offered by the displays ys a scr screen eensho shott of the Yahoo! application.   Fig. 1   displa Avatarss customisatio Avatar customisation n tool. Part Partici icipa pant ntss we were re pres presen ented ted wit with h tw two o scen scenar arios ios fo forr creatin cre ating g an avatar avatar which which depict depicted ed events events the they y had most likely encountered in the past i.e. a romantic date and a birthday postcard to a family member. Additionally, the scenarios were built on the grounds of previous research in social media. For example, bloggers often write posts with the purpose to record their life happenings (Herring ( Herring et al., 2004;;  Nardi et al., 2004 2004 2004). ). Even though blogs are frequently tar targete geted d to an aud audien ience, ce, sometim sometimes es writing writing is a solitar solitary y task, a form of ‘‘catharsis’’ for working out one’s emotions. Problems that occupy the user in daily life are expressed on onli line ne in orde orderr to reso resolve lve and to relie relieve ve the the so sour urce ce of  er and Calvert, 2005). 2005). To capture these emotion (Huffak (Huffaker

Table 1 Social media scenarios

Scenario  1: You are going out on a first and much anticipated romantic date. There are many questions you have to plan for. Where should we meet? What should I wear? How friendly should I look? What should I talk about? You decide to display your date online via your Yahoo! avatar Scenario  2: Your mom’s birthday is coming up and you have decided to surprise her with a holiday at a Greek island. You have orchestrated all the details: the perfect season for good weather, delicious local food guaranteed and a quiet hotel room by the seashore. Instead of telling her in person, you want to surprise her by presenting your vacation offer in a more fun and unexpected ‘packaging’. You will send a postcard with Yahoo! Avatars

dy dyna namic mics, s, th thee fir first st sc scen enar ario io pr promp ompted ted pa part rtici icipa pant ntss to expres exp resss a particu particular lar life event via the their ir avatar. avatar. Furthe Furtherrmore, mor e, soc social ial media media oft often en lin links ks partici participan pants ts to familia familiarr offline contacts. For instance, communicating and reconnectin nec ting g with friends friends is a motiva motivating ting facto factorr for using the social networking site Facebook (Joinson, (Joinson, 2008 2008). ). Similarly, many man y blogge bloggers’ rs’ pages pages ser serve ve as a channe channell for keeping keeping in touch with their loved ones (Nardi ( Nardi et al., 2004). 2004). This social ca capa paci city ty of so socia ciall medi media a wa wass capt captur ured ed by re requ ques estin ting g participants to construct a message for a family member. 1  displays the two scenarios. Table 1 displays Participants Partic ipants took 8.8 min on average to complete the first tas task k and spent spent an average average of 7.6 min on the second second task. The minimum minimum time spent on a tas task k was 3.4 min min,, while while the maximu max imum m was 17.45 min min.. A paired paired   t-test on task completion time was non-significant. Therefore, participants spent about equal time on both tasks.

3.3. Qualitati Qualitative ve protoc protocol ol analysis During Dur ing the thi think nk aloud aloud ses sessio sion, n, partic participa ipants’ nts’ face face and voice was recorded, as was their onscreen activity. The two ob obse serv rver erss of th thee se sess ssion ionss re revi visit sited ed th thes esee re reco cord rding ingss to identif ide ntify y mom moments ents rel releva evant nt to partici participan pants’ ts’ chosen chosen selfselfpresen pre sentati tation on str strateg ategies. ies. The contex contextt of the int intera eractio ction, n, participants’ partic ipants’ verbal reports reports and participants’ participants’ non-verbal expressions, e.g. smiling, were all consulted when extracting relevant incidents. Usability problems with the interface, e.g. this button does not work, were discarded. This process resulted in a total of 202 incidents which were then transcribed and translated in English. Next, the incidents we were re gr grou oupe ped d by si simi mila lari rity ty form formin ing g fo four ur hi high gh-l -lev evel el categories of self-presentation strategies. These are: stable self-attributes, idealisation, messages to others and fun. In the next section, we analyse the results by categorising the incidents according to this classification. 4. Resul Results ts

4.1. Stable Stable self-a self-attrib ttributes utes

Fig. 1. Yahoo! Avatars custom customisation isation tool.

Partic Par ticipa ipants nts expres expressed sed a number number of sta stable ble sel self-as f-aspec pects ts that that we were re no nott in influ fluen ence ced d by th thee two sc scen enar arios ios.. Thes Thesee

 

ARTIC TICLE LE IN PR PRESS ESS AR A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

were: (1) personalising a face, (2) fashion statements, (3) values and (4) real life events.

4.1.1. Perso 4.1.1. Personalis nalising ing a face Yaho Ya hoo! o! Av Avat atar arss pr prov ovid idee eigh eightt face face choi choice ces, s, each each of  which whi ch displa displays ys dif differ ferent ent cha charac racter teristi istics. cs. General Generally, ly, the accurate reflection of oneself via a face was an important choice cho ice,, and, and, thu thus, s, one worth the time inves investing ting in. Fiv Fivee participants chose a face at the onset, only to revisit the eight faces later, in search of a closer match to their actual visage. Unlike other avatar items that were malleable and context-driv conte xt-driven, en, once chosen, the face remaine remained d consta constant nt acro across ss the the tw two o scen scenar arios ios.. Wh When en appr approa oach chin ing g a face face,, partici par ticipan pants ts eit either her referen referenced ced the themse mselve lvess or con conside sidered red the avatar as an external agent with a character of its own. We consider each strategy separately. People define themselves in different ways, which in turn can be pro project jected ed onto onto their their ava avatar. tar. For exa example mple,, one’s one’s identity can be linked to a particular facial characteristic. A testament to our ability to recognise distinctive features and use them as identity cues are the range of adjectives we use to ‘‘classify’’ people we know e.g. he has an oval face, sh shee ha hass al alab abas aste terr skin skin.. To th that at ef effe fect ct,, on onee of ou ourr participants, while selecting a face, said:

 

805

meaning (see   Fig. 2). 2). In the following extract, one of our participants, a Caucasian female, chose an Asian face on the basis of its mood. The avatar was perceived as happy and thus reflected her state of mind, even though it did not accurately represent her own physical appearance. faces es yfac

are dif differe ferent, nt, with with dif differe ferent nt emotion emotions; s; I will maybe choose a face with a little smiley

In th thee exam example pless give given n so fa far, r, pa part rtici icipa pant ntss so soug ught ht to express their own perceived express perceived qualities and characteristi characteristics cs through their avatar. Hence, the avatar’s face should reflect certai cer tain n aspects aspects of its owner. owner. However However,, people people oft often en als also o associate personality traits to a face in the absence of any other oth er cues cues (Taylor Taylor,, 2002 2002). ). For ins instan tance, ce, one par partic ticipa ipant nt  jokingly described the avatars’ faces thus: ‘‘well-behaved’’ d’’ (laughs); (laughs); this yone avatar looks rather ‘‘well-behave one looks like a playboy (laughs) y In this context, avatar faces were treated as independent agen agents ts wi with th a di disti stinc nctt pe perso rsona nalit lity y and and co corr rres espo pond nding ing behaviours. In the above example, the participant’s final choice cho ice was to choose choose an avatar avatar face that that res resemb embled led his

This participant defined himself by the uniqueness of his almond-shaped eyes. He attributed the lack of resemblance be betwe tween en himse himself lf and and hi hiss avat avatar ar to the the misma mismatc tch h in eye eye shape completing his choice with a feeling of disappointment. Participants following this style looked for defining person per sonal al featur features, es, such as a particu particular lar facial shape or a beard they wore. The feeling resulting from one’s inability to accurately accurately represe represent nt his/herself his/herself was disap disappointme pointment. nt. Alterna Alte rnative tively, ly, ide identit ntity y may instea instead d be linked linked to one’s one’s

own.participants Thus, one strategy does not preclude and a few vacillated from one strategyanother to the other. In additi addition on to the faces faces availab available, le, partic participa ipants nts could could personalise other facial attributes such as the eye colour, hairst hai rstyle yle and colour colour,, and skin skin tone. tone. Unlike Unlike faces, faces, these these additional attributes were only partially enduring. Participants pan ts reflect reflected ed their their natura naturall eye and hair colour colour,, rarely rarely revising revisi ng these two choices. A similar observation observation was made for the skin tone as well as the hairstyle. For instance, as one participant noted ‘‘I will first select the colour of the skin. I have light skin so I will keep it this way.’’ However, during dur ing our analy analysis, sis, we als also o discove discovered red that skin skin tone tone and and hair hairst style yle we were re of ofte ten n malle malleab able le fe feat atur ures es dr driv iven en by stra strate tegic gic motiv motivat ation ions. s. This This pa part rtic icul ular ar ob obse serv rvat atio ion n is analysed in-depth in Sections 4.2 and 4.3.

un unde ders rsta tand ndin ing g of hi his/ s/he herr own own char charac acter teris istic tics. s. Here Here,, mood and mood and disp dispos osit ition ion as expr expres esse sed d by the the avat avatar ar face face became bec ame mor moree import important ant than than the similar similarity ity with partic particiipants’ physical facial features. Although the eight avatar faces fac es availab available le were given given number numbered ed labels labels by Yahoo! Yahoo! design des igners, ers, som somee par partic ticipa ipants nts ass assign igned ed them them emo emotio tional nal

4.1.2. Fashi 4.1.2. Fashion on statements particu4.1.2.1. Clothes express personal personality. ity.   People have particula larr id idea eass when when it co come mess to cons consid ider erin ing g th thee ki kind nd of  pe perso rson n th they ey ar are. e. One One may may th thin ink k of on ones esel elff as a matu mature re person, while another may want to project youthfulness.

y   well,

I do not find eyes which resemble mine y   The face does not resemble too much my own (sighs) but it will doy

Fig. 2. Male and female avata avatarr faces.

 

ARTIC LE IN PR PRESS ESS ARTICLE 806

A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

A partic participa ipant nt wearing wearing eye wear wear in rea reall life searc searched hed for glasse gla sses. s. He fro frowne wned d when when vie viewing wing the option optionss ava availab ilable le noting the disparity to his self-image: self-image: ‘‘No, the glasses glasses look to too o youn young. g.’’ ’’ An Anot other her part partici icipa pant nt wa wass unsu unsucc cces essfu sfull in finding find ing glasses glasses to match his real real life glas glasss frame. frame. As he explained, ‘‘Here it could be good to vary more, to have va vari riou ouss ty type pess of gl glas asss fr fram ames es.’ .’’’ In both both in inst stan ance ces, s, participants opted out of the eye wear option. In a similar

turned his att turned attenti ention on to the backgr backgroun ounds ds availab available le and searched again with the same purpose: ‘‘On the other hand, I hav havee the impression impression that those those are backgrou backgrounds nds so if  th ther eree is one of Swit Switze zerl rla and it co coul uld d be ni nice ce!’ !’’’ Th This is partic par ticipa ipant nt wanted wanted to expres expresss his cultur cultural al belong belonging ing via his ava avatar tar.. Whe When n realis realising ing there there were were no Swiss Swiss option optionss available, availa ble, he abandoned abandoned his search with disappointme disappointment. nt. Anot An othe her, r, Mexi Mexica can n pa part rtic icip ipan antt drew drew on hi hiss cu cult ltur ural al

vein vein,, while whto ileinclude one one pa part rtic icip ipan antt em embr brac aced ed th the e de desig ners’ s’ decision male jewellery allowing him tosigner reflect a real real life pierci piercing, ng, anothe anotherr par partici ticipan pantt cri critici ticised sed male male  jewellery as being ‘‘too feminine’’.

upbring upb ringing ing whe when n convey con veying ing the surpris sur prise e ose holida hol tospla his mo mom; m; in inst stea ead d of a Gree Gr eek k is isla land nd, , he ch chos e iday toy di disp lay y Acapulco.

4.1.2.2. Combinati 4.1.2.2. Combinations ons should should reflect good taste.   Clothes, shoes and accessories were chosen to match in both style and colour. As in real life, when choosing combinations, participants were careful to reflect good taste. Exemplifying this, one of our participants exclaimed in disgust upon try tryin ing g on a bl blue ue Hawa Hawaiia iian n sh shirt irt and and bl blac ack k Hawai Hawaiia ian n trous tro users ers ‘‘Yuc ‘‘Yuck! k! (lau (laugh ghs) s) That That does does no nott match match well. well.’’ ’’ Another participant who was looking for glasses to match with his outfit said: yI would like to find something which will correspond well well to wh what at I am wear wearin ing gy   Glasses, Glasses, bla black ck frames frames (searching) therefore I should choose this.

Th This is part partic icip ipan antt fe felt lt co comp mpel elle led d to ta take ke the the mo most st complimentary option. His choice was motivated by style and, thus, glasses that were incongruent with his outfit were not favourable. ‘‘I am a sho shoe pe pers rson on’’ ’’;; 4.1.2.3. ‘‘Got-to4.1.2.3. ‘‘Got-to-have’ have’’’ items. items.   ‘‘I ‘‘Earrings are my trademark’’. Statements of passion about particular clothing items are expressed all the time. Fashion favourites were sought after when creating online avatars. One participant, a self-reported shoe lover, exclaimed:

Yahoo! offers a bran branded ded items 4.1.3.2. Moral principl 4.1.3.2. principles. es.   Yahoo! secti section on comp compris rised ed by clo cloth thin ing g and and acces accesso sorie riess fr from om popular brands, which can be used to dress one’s avatar. This new genera generation tion of advert advertisin ising g is incons inconspicu picuous ously ly embedded in the avatar environment (Hemp, (Hemp, 2006 2006), ), but it do does es no nott come come witho without ut cr criti iticis cism. m. While While lo look okin ing g at th thee branded items, five participants expressed their opposition towa toward rdss adve advert rtis isin ing g and and fr free eedo dom m of ch choi oice ce.. As one one explained: So, this really makes me think that Yahoo! Avatars can become bec ome an adverti advertisin sing g tool tooly   prim primee peop people le to get get interested into the branded stuff. That is not something that I like much. In thi thiss instanc instance, e, partici participan pants’ ts’ ideolo ideologic gical al vie viewpo wpoint intss negatively negati vely affected their self-presentation self-presentation choices and their trust in Yahoo! As opposed to this negative outlook on online advertising, another participant cast a different view on brand advertising: ‘‘brands do not interest me, except if  they they ha have ve Deca Decathl thlon on (h (his is fa favo vour urite ite br bran and) d).’’ .’’ It is th thus us possible that inhibitions are bypassed when advertising is personalised with the inclusion of favourite offline brand names.

4.1.3.. Valu 4.1.3 Values es

4.1.3.3. Emotiona 4.1.3.3. Emotionall attach attachments. ments.   Sport Sportss fa fans ns fe ferv rven ently tly vouch for their team. Motorcycle riders take yearly group tours tou rs to show show their their solidar solidarity ity to one another. another. Hobbie Hobbiess require both a positive emotional bias and time commitment. Yahoo! Avatars provides many choices for showcasing one’s hobbies, which in turn appeared to be welcomed by participants. An ice skater chose an ice ring; a baseball fan dressed dressed up in uniform uniform;; a motorc motorcycl yclee rid rider er placed placed a moto mo torc rcyc ycle le ne next xt to his virt virtua uall se self. lf. A hock hockey ey fan fan wh who o selecte sele cted d a hockey hockey sta stadium dium backgr backgroun ound d sai said d ‘‘g ‘‘good ood,, they they have hockey because I practise it; I adore hockey (laughs). Maki Ma king ng th this is choic choicee sh show owss a li littl ttlee of my pers person onali ality ty.’’ .’’ Anothe Ano therr object object of emotion emotional al att attach achmen mentt was animals animals.. Animals require time commitment and constant care, but in return they offer us friendship and companionship. We attribute personality to animals which we learn to accept

San n rldw Fran Fr anci cisco Ve Veni nice cens and an d 4.1.3 4.1.3.1. .1. affili Mosc Mo scow owGroup ar aree a affiliation few fe wations. of s.the th e Sa worl wo dwid ide e sco, dest de,stin inat atio ions on disp display lay.. In the the follo followi wing ng exam exampl ple, e, a part partici icipa pant nt first first searched for Swiss clothes. He said ‘‘doubt that there are Swiss Sw iss clo cloth thes, es, wh which ich is quite quite a pity pity!’’ !’’ Sh Short ortly ly af after ter,, he

and love. this emotional bond wastheir also evident inThe thisimportance study whenof several participants chose real life companions as their online companions. As one partic par ticipa ipant nt noted noted ‘‘I am lookin looking g for a dog that actual actually ly look lo okss li like ke my do dog g wh whic ich h is a Chih Chihua uahu hua. a.’’ ’’ Anot Anothe herr

Are there shoes? Is my avatar wearing shoes? Can one change the shoes?y Because shoes are important to me. But, there seems to be no such option (sigh). In Yaho Yahoo! o! Avat Avatar ars, s, sh shoe oess come come auto automat matic icall ally y wit with h corresponding clothes and are one of the few options that are not customi customisab sable. le. Yet, thi thiss partici participan pant’s t’s disapp disappoin ointtment was echoed throughout the study by eight others who repeate rep eatedly dly search searched ed for ways ways to cus custom tomise ise their their sho shoes. es. Crucia Cru cially lly,, a few few part partic icip ipan ants ts canc cancel elled led thei theirr clot clothin hing g option opt ionss whe when n the corres correspon pondin ding g sho shoes es wer weree not to the their ir liking. In a similar vein, some female participants looked for makeup makeup,, a male male partic participa ipant nt search searched ed for a pie pierci rcing ng similar to his own.

 

ARTIC LE IN PR PRESS ESS ARTICLE A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

participant selected a cat when customising her avatar for the birthday scenario as she always takes her cat with her when on holiday.

4.1.4. Real life even 4.1.4. events ts Customisation options present in the site were used to express real life occurrences. For example, a few participants chose darker skin tone to reflect the summer season. Another participant expressed his recent visit to the barber with the corresponding avatar option: ‘‘Good, today I went to the hairdresser, therefore, I choose this one: very short br brow own n ha hair ir.’’ .’’ Thou Though gh the the stud study y to took ok plac placee duri during ng the the summer, when attending on a rainy day, one participant chose long pants and a hat to keep himself protected from the weather. weather.

4.2. Ideal Idealisati isation on In   Goffman’s (1959 (1959))   terms, parti participants cipants idealised their self-presentation by intentionally emphasising or concealing self-attributes. self-attributes. We found three patterns of idealisation: idealisation: (1) (1) gl glob obal al id idea ealis lisati ation on,, (2) (2) soc socia iall id idea ealis lisati ation on and and (3) (3) idealisation versus breaking the rules.

4.2.1. Global id idealisation ealisation Severa Sev erall par partic ticipa ipants nts slight slightly ly mor morphe phed d their their ava avatar tar to 1987). ). display an ‘‘ideal’’ and more attractive self (Higgins, ( Higgins, 1987 Fo Forr exam exampl ple, e, av avat atar arss were were gi give ven n a dark darker er skin skin tone tone expressing their owners’ desire for a more vibrant summer look. One participant acknowledged this embellishment by saying ‘‘I will pretend to look more suntanned.’’ Furthermore, a male participant with a receding hair line in real life started off with the intention to accurately reflect his hairstyle. Instead, he was confronted with hair options at two ends of a spectrum, either a completely bald head or a head with short hair. He negotiated these two possibilities: Let’s Let ’s try so some methi thing ng with with no ha hair, ir, espe especia cially lly in the the forehe for ehead. ad. That That loo looks ks clo close se (wh (while ile vie viewin wing g the sho short rt ha hair ir). ). Not Not as much much,, ma mayb ybee in a few few year yearss fro from m now now (laughs) (while viewing the bald head). Even tho Even though ugh thi thiss partici participan pant’s t’s ini initia tiall int intent ention ion was to truthfully depict his appearance, he found himself caught between betwee n two extremes: extremes: his past self and possibly future self. His final choice depicted an idealised and more youthful appear app earanc ance, e, chiefly chiefly impose imposed d by the limited limited hair hair cho choice icess offered within the application.

4.2.2. Social ideal idealisation isation scenario, partiparti4.2.2.1. 4.2. 2.1. Roma Romantic ntic partner. partner.   In the dating scenario, cipants predominantly enacted the role they usually play when impressing a romantic partner. Setting, appearance or was accordingly. In some extreme al alll manner thre threee fa face cets tsaltered of fr fron ont t were were used used in co conc ncer ertt for focases, r the the performance perfo rmance.. Specifically, Specifically, many participants participants chose backgrounds of conventional locations for a romantic meeting e.g. a dinner evening in a terrace, a date at a disco. Others

 

807

searched search ed for a set setting ting that depict depicted ed partic particula ularr tra traits. its. To givee an example giv example,, one partici participan pantt selected selected a ‘‘tranq ‘‘tranquil’ uil’’’ hiking hik ing locatio location, n, jus justt right right for getting getting to know know someon someonee new.. Partic new Participa ipants nts search searched ed for elegant elegant clo clothe thes: s: severa severall female fem ale par partic ticipa ipants nts selecte selected d evenin evening g dresses dresses,, while while male male participants selected suits or formal pants. Other props for building up an attractive self-presentation were used such as jewellery, handbags or makeup. Finally, a few participants selected a evening. happy expression for their avatar, suitable for a romantic Unlike global idealisation which was guided guided by one’s one’s person personal al expect expectatio ations ns (se (seee Section Section 4.2.1), in this instance an ‘‘ideal’’ self was oriented to the expectations of a romantic partner. thee bi birt rthd hday ay 4. 4.2. 2.2. 2.2. 2. Broa Broadc dcast aster er of happ happy y news news..   In th scenario, a few participants morphed their self-presentation to align to the role of the ‘‘happy broadcaster’’. A male partici par ticipan pantt chose chose a guitar guitar to announ announce ce the happy happy news: news: ‘‘I will choose the avatar with the guitar. I could maybe sing something for my mom; happy birthday, of course!’’ A fem female ale partic participa ipant nt took took the role of the messenger messenger by dressing her avatar with festive clothes accompanied by a happy emotion expression: I am supposed to announce the gift so I need to wear something special (chooses a skirt). There we go, warm colours (chooses an orange shirt). Happy, sad, special (browsing through the avatar’s emotions). We will go with happy. It’s perfect. One pa One part rtic icip ipan antt ha had d br brow owse sed d th thro roug ugh h th thee avai availa labl blee emotion expressions, while completing the dating scenario. He said: ‘‘special, what is the special special emotion? O no! This is lousy, it is a loser’s look y   Sad emotion, no not sad y Happy emotion, this is not happy, it is idiotically happy.’’ However, when personalising the avatar with his mom in mind, he noted ‘‘since this is for my mom, I will put up a smile even though I will look stupid.’’ Though the avatar’s emotion emo tionss were were not ali aligne gned d to his expres expressive sive prefere preference nces, s, when whe n announ announcing cing happy happy news news he abando abandoned ned his inhibi inhibi-tions and selected a happy expression. completing the birthday birthday scenario, 4.2.2.3. Child.   While completing so some me pa part rtic icip ipan ants ts dr drew ew on th thee ro role le of th thee ch chil ild d who who is ob oblig liged ed to conf confor orm m to an ‘‘o ‘‘oug ught’ ht’’’ se self lf (Higgins Higgins,, 1987 1987). ). Specifi Spe cificc featur features es in their their appear appearanc ancee were were concea concealed led to align to their parent’s expectations. For instance, a female participant browsed through several swimming suits before choosing the most conservative ‘‘less naked’’ choice. In the following follow ing extract, extract, a participant participant who had enthusiastically enthusiastically chosen a piercing similar to his own for the date scenario removed the piercing while explaining: ‘‘I will take off the piercing; otherwise my mom will tell me that I am ‘scum’ (laughs).’’

4.2.3. Ideali 4.2.3. Idealisatio sation n versu versuss breaking breaking the rules Even though most participants drew on existing social roles roles during during the proces processs of customi customisat sation ion,, avatar avatarss als also o

 

ARTIC LE IN PR PRESS ESS ARTICLE 808

A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

encouraged ‘‘breaking the rules’’ as contrasting items could be easily combined into one composition with as little as one click. A female participant participant chose ‘‘sophi ‘‘sophisticate sticated’’ d’’ Paris as her destination, while wearing a cowboy outfit she had earlier selected for her date. She hesitated for a moment but then the n emb embrac raced ed her ‘‘inapp ‘‘inapprop ropria riate te cho choice’ ice’’: ’: ‘‘h ‘‘here ere we go (select (sel ected ed Par Paris); is); am I wea wearing ring clothing clothing suited suited for dow downntown? Not really, actually, (pause) but well, I like Paris, let

4.3.2. Provoking 4.3.2. Provoking a reaction Severall participants Severa participants explicitly morphed their avatar avatar to provoke provo ke an emotional reaction from the message recipient. In the following two examples, the sender envisioned the receiver’s response, experiencing positive empathy for the other oth er and assigning assigning even more more emotion emotional al value value to their their cu cust stom omis isati ation on choi choice ce.. A pa part rtici icipa pant nt who who had had earl earlier ier selected a self-representative hairstyle now said ‘‘I would

us go! (laughs). (laughs).’’ ’’ A male male par partici ticipan pantt who had chosen chosen a ‘‘chic’ ‘‘ch ic’’’ suit suit purpos purposefu efully lly for the date, date, browse browsed d thr throug ough h backgr bac kgroun ounds ds in search search of a romant romantic ic locatio location. n. His fina finall choice was a farm. He noted the incongruence between his idea idealis lised ed appe appear aran ance ce and and the the back backgr grou ound nd:: ‘‘t ‘‘the he farm farm background is not really aligned to the suit (and thus to th thee date date), ), bu butt it match matches es the the sty style le of the the ha hat.’ t.’’’ Desp Despite ite setti se tting ng ou outt to pe pers rson onif ify y the the ‘‘ro ‘‘roma mant ntic ic part partne ner’ r’’, ’, this this pa part rtic icip ipant ant wa wass able able to quic quickl kly y and and easi easily ly di dism smis isss the the social conventions he had initially chosen to follow.

like to change the haircut for the beach. I want to find something funny for my momy   Afro hairy   This one is funny (smiles).’’ In a similar vein, a second participant gave hi hiss avat avatar ar a su suit it even even th thou ough gh hi hiss tast tastee in fa fash shio ion n was was casual. He explained his mom’s reaction if she were to view his avatar: ‘‘She will be proud of the little boyy My avatar is smartly dressed; that will give her a shock (laughs).’’

4.3. Messag Messages es to others 4.3.1.. Liter 4.3.1 Literal al versus symbo symbolic lic messages In the birthd birthday ay scenario, similar to what was observed in the social idealisation strategy (see Section 4.2.2) participants orchestrated setting, appearance and manner during the avatar construction construction.. However, However, instea instead d of adapting adapting a pa part rtic icula ularr so soci cial al ro role le or orie iente nted d towa toward rdss an imag imagin ined ed audience, audien ce, all but two participants participants enacted their ‘‘model’’ beach bea ch holida holiday. y. Theref Therefore ore,, the ava avatar’ tar’ss app appear earanc ancee was mani ma nipu pulat lated ed to send send a lite litera rall messa message ge to a pa part rticu icula larr re reci cipi pien ent. t. Be Beac ach h back backgr grou ound ndss were were ta take ken n that that were were indi indicat cativ ivee of a Gr Gree eek k islan island. d. Avat Avatar arss we were re dr dres esse sed d in bathin bat hing g suits suits or in light light summer summer gar garmen ments. ts. Props Props were were used use d to fur further ther hig highlig hlight ht the holida holiday: y: some some partici participan pants ts chos chosee su sun n gl glas asse ses; s; ot othe hers rs chos chosee su summe mmerr hats hats;; and and two two fe femal malee pa part rtici icipa pant ntss chos chosee su summe mmerr ha hand ndba bags gs.. Se Seve vera rall partic par ticipa ipants nts col colour oured ed the their ir skin ton tonee darker darker to furthe furtherr illus illustra trate te the the beac beach h holi holida day, y, where whereas as one one part partici icipa pant nt lighten ligh tened ed her skin tone to expres expresss the fact that she had not yet left for the holiday. Finally, two female participants chose cho se more more casual casual hairsty hairstyles les suit suitabl ablee for the beach. beach. As one explained explained:: ‘‘I would like a haircu haircutt for the hol holida idays. ys. The The sm smal alll po pony nyta tail il is no lo long nger er fit. fit. Oh, Oh, dr drea eadl dloc ocks ks are are good.’’ Ev Even en thou though gh av avat atar arss were were pred predom omin inan antl tly y used used to communicate literal messages, the abundant visual options with within in the the site site al also so le led d to the the co cons nstr truc ucti tion on of some some symbolic messages. One participant used imagery to attach informat info rmation ion of sign significa ificance nce to the messag messagee recipie recipient. nt. He said ‘‘my mom likes animals a lot and cats in particular.’’ In orienting his avatar towards another, he searched for a cat companion. Another participant used the background image to construct a metaphor: a black background filled with light bulbs communicated her brilliant holiday idea. Yet, a third participant chose a birthday party background with a cake and gifts to signify the ‘‘festiv ‘‘festive’’ e’’ birthday spirit.

4.4.. Fun 4.4 Finally,, some participants Finally participants gravitated gravitated towards towards options options that were simply fun and entertaining to them personally. One participant chose an amusing hat because it reminded him of ‘‘Indiana Jones’’. Another participant chose a dog simi simila larr to her her own own whic which h was was dres dresse sed d in a rain rainco coat at.. Laughi Lau ghingl ngly, y, she commen commented ted twice twice on her dog’s dog’s comica comicall appear app earanc ance. e. Many Many oth others ers assign assigned ed their their avatar avatar an anianimated emotion because of its entertainment value. In the followi foll owing ng exampl example, e, a partici participan pantt browsed browsed throug through h the emot em otio ions ns ‘‘ ‘‘sa sad’ d’’, ’, ‘‘ ‘‘an angr gry’ y’’’ an and d ‘‘ ‘‘sp spec ecia ial’ l’’, ’, in th thee end end selecting the ‘‘special’’ emotion: Excellent! Yes! (selects the sad emotion) Ohhhh, he is sad, he is very sad (laughs). (selects the angry emotion) This one is angry, not bad. (selects the special emotion) And the special one, yes the special one, it is cool the special one (laughs). Let us zoom into see, do again that littl littlee th thin ing g you you di did d earl earlier ier,, yeah yeah,, he is cool cool my guy guy (laugh)y 5. Discu Discussion ssion

The discussion that follows organises the self-presentation strategies delineated delineated in Section Section 4 into three motivating factors driving self-presentation: (1) accurate self-presentation, (2) playfu playfull self-present self-presentation ation and (3) embodied messages. We acknowledge that the three motivating factors do not necess nec essaril arily y ope operat ratee indepe independe ndentl ntly; y; while while an avatar avatar user user may ma y re refle flect ct an accu accura rate te se self, lf, ele eleme ment ntss of pl play ayful ful se selflfpresentation may be additionally chosen.

5.1. Accurate Accurate self-presen self-presentation tation:: mirro mirroring ring one’s offlin offlinee self  onto an avatar Participants reflected their ( Higgins, (Higgins, 1987 1987) ) bysele selecti cting ng option options s ali aligne gned d to actual their their self percei perceived ved sta stable ble selfself attributes. attribu tes. Across both scenarios, scenarios, avatars’ avatars’ physical physical characteris act eristic ticss were were chosen chosen careful carefully ly to reflect reflect their their owners’ owners’ appearance. Fashion preferences that guided participants’

 

ARTIC LE IN PR PRESS ESS ARTICLE A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

 

809

appearanc appear ancee in the physica physicall wor world, ld, also det determi ermined ned the their ir online onl ine cho choices ices.. Moreov Moreover, er, par partici ticipan pants ts dre drew w on a larger larger framework of stable values when personalising their avatar including cultural affiliations and moral beliefs. In add additio ition n to expres expressing sing an act actual ual sel self, f, par partici ticipan pants ts idea id ealis lised ed their their avat avatar ar by conc concea ealin ling g or emph emphas asisi ising ng particu par ticular lar self-at self-attrib tribute utes. s. Firs First, t, soc social ial role roless det determ ermine ined d how partici participan pants ts cho chose se to presen presentt the themse mselve lvess (Goffman,

the the fo form rm of id idea ealis lisat atio ion. n. To achi achiev evee th this is,, parti partici cipa pants nts we welc lcom omed ed th thee sp spec ectru trum m of choi choice cess made made avai availa labl blee by Yahoo! Yah oo! design designers ers for set setting ting,, appear appearanc ancee and manner manner.. Hobb Ho bbie ies, s, pets pets and and ot othe herr sel selff-rel relev evan antt pr prop opss we were re all all import imp ortant ant ele element mentss for identi identity ty constru constructio ction n (Goffman, 1959). 1959 ). Thus, alongside the control offered over one’s online embo em bodi dimen ment, t, desig designe ners rs of avat avatar ar cu cust stom omis isat atio ion n to tool olss sh shou ould ld put put fo forw rwar ard d othe otherr arte artefa fact ctss th that at are are equa equall lly y

1959). When expressing an impending romantic date, many 1959). users use rs constru constructe cted d avatars avatars that that loo looked ked mor moree attr attract active, ive, while backgrounds depicted popular romantic destinations. Ther Theref efor ore, e, an idea ideall self self wa wass pr prev eval alen ent, t, comp compos osed ed of  qualities desired by a romantic partner. When constructing a message for their mom, a few participants projected an ought self (Higgins, (Higgins, 1987) 1987) by intentionally creating a more conservativ conse rvativee appea appearance. rance. Some participants participants also adop adopted ted the role of the broadcaster, whose aim was to deliver good news with the display of a happy and festive fac ade. When participants constructed the birthday surprise, even though the avatar itself constituted the message (see Section 5.3), they they drew drew toge togethe therr al alll th thee ele eleme ment ntss that that desc describ ribed ed a holidaymake holid aymaker. r. Second, Second, some participants participants idealised their avatarr with slight improv avata improvements ements they person personally ally desired

important to identity important identity expression. expression. Furthermore, Furthe rmore, when constructing constructing one’s self-presenta self-presentation, tion, lack of flexibility over one’s options ultimately determined the the avat avatar’ ar’ss fin final al appe appear aran ance ce and, and, th thus us,, th thee de degr gree ee of  idealisation the avatar owner expressed. This latter point has some implications implications for interpersona interpersonall communication communication.. Yee and Bailens Bailenson on (2007)   found found that participants participants represented by attractive avatars were more willing to approach members of the opposite gender, as compared to participants represented by less attractive avatars. In a negotiation tion ta task sk,, part partici icipa pant ntss with with ta tall ller er avat avatar arss we were re mo more re confident than participants represented by shorter avatars. By preventing or enabling the display of particular avatar att attrib ribute utes, s, avatar avatar design designers ers exert exert influen influence ce over over users’ users’ idealisation, ideali sation, in turn possibly shaping users’ online beha-

(Higgin Higgins, s, 198 1987 7). No None neth thel eles ess, s, this this type type of inte intent ntio iona nall embellishment was not a predominant trend in this study. As origin originally ally propo proposed sed in Sectio Section n 2.2, 2.2, we found found tha thatt pa part rtici icipa pant ntss comp compos osed ed their their avat avatar ar to giv givee of offf prec precis isee messages messag es that intentionally intentionally avoided ambiguity. A female participant chose an evening dress, to later notice a nonrepresentative music sheet in the hands of her avatar. She opted for another dress even though it was less congruent to her fashio fashion n style. Similarly, Similarly, a male participant participant searching for a suit aimed to display elegance without ‘‘diluting’’ his self-image self-im age with other attributes: ‘‘I will choos choosee this suit as it is more more eleg elegan ant. t. The The ot othe herr one one look lookss too too mu much ch like like a businessman busin essman.’’ .’’ Howeve However, r, even thoug though h participants participants carefully fully selecte selected d option optionss that that were aligne aligned d to an ima imagin gined ed social role, they tended to be reflective of their stable selfatt attrib ribute utess and prefer preference encess (see (see Sec Sectio tion n 4.1). 4.1). As a fem female ale participant stated: ‘‘the first date is important for me. I will try to find a way of dressing nicely, but that also reflects me. I am looking for something that looks like a student. Th This is one one may may be too too yo youn ung. g. I woul would d like like so some meth thin ing g pleasanty  This one reflects me so I save.’’ Precise control over one’s idealised self-presentation and the need for selfrepresentative options were competing factors that directed partici par ticipan pants’ ts’ choice choicess eit either her closer to an act actual ual sel selff or to strate strategies gies of idealis idealisati ation. on. In an illustra illustrative tive exa exampl mple, e, one participant had earlier chosen glasses with a frame similar to his own. own. When When comple completing ting the birthd birthday ay sce scenar nario, io, he looked for sunglasses. He searched in vein for a pair he liked, finally deciding to keep his original eye wear.

viour (also see Taylor, see  Taylor, 2003). 2003).

In Sect Sectio ion n 2. 2.3, 3, it was was sugg sugges este ted d th that at some some avat avatar ar customisation tools present equal availability and prominence nen ce of fictitio fictitious us objects objects and als also o objects objects found in the physical world. This introduces a level of ambiguity that in turn challenges one’s own self-conception, but also one’s social orientation towards another. Aligned to this prediction tion,, we fo foun und d th that at th thee ambi ambigu guity ity fo fost ster ered ed in Yaho Yahoo! o! Avat Av atar arss le led d to fiv fivee ty type pess of st stra rate tegi gies es.. Part Partic icip ipan ants ts abandoned aband oned their preconceptio preconception n of social correctness correctness and experimented exper imented by mixing ‘‘inappropriate’’ ‘‘inappropriate’’ choices choices together. together. Theref The refore ore,, social social rules rules that that ste steer er one’s one’s self-pr self-prese esentat ntation ion offline offl ine were were bent bent and circumv circumvente ented d online online.. Partic Participa ipants nts also used the avatar to build narratives about events taking place in their physical world. As opposed to reporting one’s present prese nt state via a textual channel, as offered in most social media platforms, e.g. blogs, social networking applications, part partic icip ipan ants ts used used the the avat avatar ar’s ’s body body as a me mean anss fo forr expres exp ressing sing their their ongoin ongoing g life events events.. Furthe Furthermor rmore, e, some some constru con structe cted d symboli symbolicc and meta metapho phoric rical al messag messages es that that invited invite d the viewer’s interpretation. interpretation. Additionally, Additionally, participants manipulated their appearance in order to provoke a positive emotional reaction from the avatar viewer. Finally, sele select ctio ions ns we were re made made th that at we were re no nonn-re repr pres esen enta tativ tivee of  participants’ offline self, but were chosen because of their

Overal l,a the above con clusio sion n is line linale med withiaobserv observaa-t tional tioOve nal rall, dat data sug sugges gesting tingconclu that use users rs of in soci social media presen present themselves themse lves accura accurately tely (Herring et al., 2004 2004;;  Vasalou et al., 2007). 2007 ). In thi thiss study, study, acc accura urate te self self-pr -prese esenta ntation tion was bui built lt into avatars either in the form of stable self-attributes or in

entertainment As discussedvalue. in Section 5.1, many participants aimed to construct a precise self-presentation, as a result criticising the avatar application over its lack of flexibility. However, the presence presence and diversity diversity of predefi predefined ned (and thus thus les lesss

-

5.2. Play Playful ful self-presen self-presentation tation:: reinve reinventin nting g onesel oneselff and social  life

 

ARTIC TICLE LE IN PR PRESS ESS AR 810

A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

customisable) objects also created a new space for creative self-ex self -expre pressio ssion n and commun communicat ication ion.. Par Partici ticipan pants ts who exploited the ambiguity afforded by the avatar customisation tool advanced their identity beyond the mere reflection of an offline self. The present result is also in line with obse observ rvat atio ions ns from from virt virtua uall en envi viro ronm nmen ents ts;; to gi give ve an illustrative example, in a study reported by   Taylor (2002), (2002), a female avatar user distanced herself from users she was

consequence, it is likely that embodied messages facilitate intense emotional exchanges between message sender and re rece ceiv iver er,, ev even en more more so th than an th thee ex exch chan ange ge of fleet fleetin ing g emotion expressions.

not familiar familiar with, with, symboli symbolicall cally, y, by wea wearing ring a witc witch h hat hat.. Hence, eve Hence, even n tho though ugh com comput putatio ational nal res resour ources ces may ultiultimately allow users to shape their online representation with precisi pre cision, on, a strong strong arg argume ument nt can be made made in fav favour our of  ambigu amb iguity ity and the use userr engagem engagement ent it fos foster ters. s. We thus thus argu arguee that that de desig signe ners rs of avat avatar ar envi enviro ronme nment ntss sh shou ould ld cont contin inue ue main mainta tain ining ing a pr pred edefi efine ned d palet palette te of di dive verse rse customisation options.

analysee users’ self-presentation analys self-presentation strategies when creating creating avatarss for social media. avatar media. The con concur curren rentt format format of the think thin k aloud aloud method method allowed allowed us access access to partici participan pants’ ts’ though tho ughts ts in order order to gain gain thi thiss fine-gr fine-grain ained ed knowled knowledge. ge. However, Howeve r, this methodological methodological choice choice also placed participarticipants outside their natural environment and inside the lab; the social media scenarios motivating the avatar creation cannot capture the entire spectrum of participants’ intention of use. It is important that future research corroborate ratess and and exte extend ndss ou ourr re resu sults lts by et ethn hnog ogra raph phic ic mean means. s. Moreover, even though the facilitator was trained not to interfere interfe re during during participants participants’’ avatar avatar creation, creation, we recognise recognise that the observation methods used in this study may have prevented preven ted some participants participants from verbalising verbalising particular, particular, perhapss more embarrassing perhap embarrassing,, self-presenta self-presentation tion strategies. strategies.

5.3. Embo Embodied died messages: messages: avatars as proxies proxies When announcing a surprise holiday gift on the occasion of their mom’s birthday, birthday, participan participants ts enact enacted ed their ideal holiday holida y break via their avatar. Beach backgr backgrounds ounds were preferred, while bathing suits and other holiday items were

5.4. Limitatio Limitations ns To our our know knowled ledge ge,, th this is st stud udy y is th thee fir first st atte attemp mptt to

Finally, Fina lly, Yahoo! Yahoo! fre freque quently ntly update updatess the Yahoo! Yahoo! Avatar Avatarss customisation tool with new selections. As the site grows, us user erss may may cont contin inue ue in inve venti nting ng ne new w st stra rate tegie giess fo forr se selflfpresentation.

selec se lected ted.. By us usin ing g fron front, t, pa part rtici icipa pant ntss expr expres essed sed thei theirr holiday self in a similar manner to those who engaged in social idealisation. However, in this instance, one’s digital representation embodied a literal message towards another. Researchers Resear chers and practitioners practitioners have previously previously focuse focused d on creatin creating g ava avatar tarss with inin-bui built lt emotio emotion n expres expressiv sivity ity in order to enable the exchange of richer emotional content. For exampl example, e, in the V-Ch V-Chat at virtua virtuall env enviro ironme nment, nt, users users were given the ability to compliment their written messages with static emotional gestures (Cheng ( Cheng et al., 2002). 2002 ).   Fabri and Moo Moore, re, 200 2005 5   built built a chat chat mes messen senger ger which which allowe allowed d users use rs to change change their their avatars avatars’’ fac facial ial emo emotio tion n exp expres ression sion.. This kind of approach naturally follows from our intrinsic communicative commun icative mechanisms. However, the present present result indicates that avatar users can invent new and unconventional communicative gestures such as using the avatar’s appearance to embody the message. Future Fut ure wor work k could could invest investigat igatee users’ users’ com commun municat icative ive intentions in order to reveal what kind of customisation opti option onss ar aree need needed ed to supp suppor ortt em embo bodi died ed me mess ssag ages es.. Additio Add itionall nally, y, fur furthe therr res resear earch ch might might examine examine whe whethe therr embodied messages influence communication at the interpe pers rson onal al lev level. el. The The expr expres essio sion n and and in inte terpr rpret etat atio ion n of  emotions is frequently socially and culturally constructed. A reciprocal emotional exchange can occur in which the send se nder er’s ’s emot emotio ion n expr expres essio sion n elici elicits ts an emot emotion ion in the the receiver (Parkinson, (Parkinson, 2008 2008). ). Aligned to this view,  Fabri and Moore (2005 (2005))   found found that users chatting chatting online mirrored each others’ emotions as expressed via their avatar. In this

The present present res resear earch ch set out to invest investigat igatee how users choose to present themselves via an avatar for social media. Using the concurrent think aloud protocol, three distinct motiva mot ivating ting fac factor torss in avatar avatar constr construct uction ion were were found. found. First, participants depicted  accurate   self-presentations that were aligned either to an actual self, as expressed by one’s perceived stable self-attributes, or to an ideal or ought self, as expr expres esse sed d by st stra rate tegi gies es of id idea ealis lisat atio ion. n. St Stab able le se selflfattributes and idealisations are competing strategies, often resolved by the availability of options within the customisat sation ion too tool. l. Avatar Avatar design designers ers may want want to acknow acknowled ledge ge users’ need for constructing precise self-representations by offering additional control and flexibility over customisation tion cho choice ices. s. Furth Further er,, while while cr crea eatin ting g an accu accura rate te se selflfpresen pre sentati tation, on, partici participan pants ts drew drew togethe togetherr a collect collection ion of  objects that depicted setting, appearance and manner. On these the se gro ground unds, s, we sugges suggestt wideni widening ng the scope scope of avatar avatar customisation custom isation to include include options options that are not necessarily attached to the avatar’s body but are yet central to users’ identity identit y (e.g. pets). Second, Second, ambiguity ambiguity allowed allowed participants participants to create  playful   self-presentations. Some participants used the avatar to challenge the social conventions that usually

study, discovered thatself senders are happy to give out up their need towe depict an actual in favour of reaching and enga engagin ging g with with anot anothe herr us user er.. Co Conv nver ertin ting g on one’s e’s di digi gital tal representation into an expressive signal may be perceived by the the mess messag agee recip recipien ientt as an altr altrui uist stic ic gest gesture ure.. As a

direct self-prese self-presentatio ntation n when Others built or narratives into the avatar’s body thatoffline. reflected life events that directly dire ctly engage engaged d the vie viewer wer.. Thus, Thus, even even though though ava avatar tar applica app licatio tions ns may hav havee the future capaci capacity ty to all allow ow the creatio cre ation n of precis precisee and pho photore toreali alistic stic represe representa ntation tions, s,

6. Concl Conclusion usion

 

ARTIC LE IN PR PRESS ESS ARTICLE A. Vasalou et al. / Int. J. Human-Computer Studies 66 (2008) 801–811

 

811

desi design gner erss shou should ld co cont ntin inue ue to supp suppor ortt am ambi bigu guit ity y by additionally addit ionally provid providing ing pre-construc pre-constructed ted customi customisation sation options. Third, avatars were used to   embody   a message as pa part rtici icipa pant ntss orie oriente nted d their their enti entire re self self-p -pres resen enta tatio tion n to convey con vey a literal literal mes messag sagee to anothe anotherr user. user. The Theref refore, ore, the av avat atar ar’s ’s body body can can be co conv nver erte ted d by the the user user in into to an expressive signal. Future research can advance the design of avatar customisation tools by gaining a deeper under-

Social Life of Avatars: Social Avatars: Presen Presence ce and Inte Interac ractio tion n in Shared Shared Virt Virtual ual Environments. Environm ents. Springer, London, pp. 90–111. Ellison, N., Heino, R., Gibbs, J., 2006. Managing Managing impressions online: selfpresentation processes in the online dating environment. Journal of  Computer-Mediated Communication 11. Fabri, M., Moore, D.J., 2005. Is empathy the key? Effective communicatio tion n via ins insta tant nt me messa ssagin ging. g. In: In: Proce Proceedi edings ngs of the 11th 11th EAT EATA A International Internat ional Conference on Networking Networking Entities. Entities. Gaver, Gav er, W.W., Beaver, J., Benford, Benford, S., 2003. 2003. Ambi Ambiguit guity y as a reso resource urce for de desig sign. n. In: In: Pro Procee ceedi dings ngs of th thee CH CHI. I. ACM Press, Press, Ne New w York, York,

stan standin ding g on the the oc occa casio sions ns that that gi give ve rise rise to em embo bodi died ed messages. However, this result also opens up new questions for online online commun communicat ication ion.. Orie Orientin nting g one one’s ’s ent entire ire body body towa towards rds anot anothe herr may may be pe perc rceiv eived ed by the the view viewer er as an altrui altruistic stic and genero generous us gestur gesture, e, in tur turn n facilit facilitatin ating g the exchange of highly emotive messages. In conclusion, the above findings advance the topic of  identit ide ntity y in social social medi media a by reveal revealing ing the strateg strategies ies tha thatt users bring together when creating an avatar. Moreover, they suggest several implications for the design of avatar customi cus tomisat sation ion too tools ls and more more genera generally lly for onl online ine com com-munication.

pp. 233–240. Goffman, E., 1959. The Presentation of Self in Everyday Life. Doubleday, Goffman, Garden City, NY. Hemp, P., 2006. Avatar-based marketing. Harvard Business Review. Herring, S.C., Scheidt, L.A., Bonus, S., Wright, E., 2004. Bridging the gap: a genre analysis of weblogs. In: Proceedings of the 37th Hawaii International Internat ional Conference on System Sciences (HICSS-37). Hess, U., 2001. The communication of emotion. In: Kaszniak, A. (Ed.), Emotions, Emotion s, Qualia, and Conscious Consciousness. ness. World Scientific Publishing, Publishing, Singapore, Singapor e, pp. 397–409. 397–409. Higgins,, E.T., 1987. Self-disc Higgins Self-discrepancy repancy theory. Psychological Review 94, 1120–1134. Huffaker, Huffak er, D.A., Calvert, S.L., 2005. Gender, identity, and language use in teenage blogs. Journal of Compute Computer-Media r-Mediated ted Communication 10. Isomursu, M., Tahti, M., Vainamo, S., Kuutti, K., 2007. Experimental evaluation evaluati on of five methods for collecting emotions in field settings with mobile mobi le applica applicatio tions. ns. Internat Internationa ionall Jour Journal nal of Hum Human–C an–Comp omputer uter Studies 65, 404–418. Joinson, A., 2008. ‘Looking at’, ‘looking up’ or ‘keeping up with’ people? Motives Moti ves and uses of faceboo facebook. k. In: Proceedi Proceedings ngs of the CHI. ACM Press, New York, pp. 1027–1036. Keltner, D., Buswell, B.N., 1997. Embarrassment: its distinct form and appeasement appeasem ent functions. functions. Psychological Psychological Bulletin 122, 250–270. 250–270. Mandryk, R.L., Atkins, M.S., 2007. A fuzzy physiological approach for continuously modeling emotion during interaction with play technologies. International Journal of Human–Computer Studies 65, 329–347. Nardi, B.A., Schiano, D.J., Gumprecht, M., Swartz, L., 2004. Why we blog. Communications of the ACM 47, 41–46. Nowak, Now ak, K.L., Rauh, C., 2005. 2005. The influen influence ce of the avatar on onli online ne perceptions of anthropomorphism, androgyny, credibility, homophily, and attraction. attraction. Journal of Computer-Mediate Computer-Mediated d Communication Communication 11. Parkinso Park inson, n, B., 2008. 2008. Emo Emotion tionss in direct direct and rem remote ote social social interact interaction ion:: getting through the spaces between us. Computers in Human Behavior 24, 1510–15 1510–1529. 29. Schroeder, Schroede r, R., 2002. Social interaction in virtual environments: key issues,

Acknowledgements

We thank Jarmo Laaksolahti (SICS) and Kristina Ho ¨ o ¨ k (KTH (KTH/S /SIC ICS) S) for for thei theirr co cons nsta tant nt supp suppor ortt du duri ring ng this this research. Etienne Roesch (University of Geneva), Sophia Vasalou (University of Cambridge) and Marc Fabri (Leeds Metropolitan Metro politan)) are acknowledged acknowledged for their comments on an ea earl rlie ierr draf draftt of this this pape paper. r. We ar aree than thankf kful ul to thre threee anon anonym ymou ouss revie reviewe wers rs wh whos osee comm commen ents ts imp impro rove ved d the the pa pape per. r. St Stev evee Bingg Binggeli eli,, De Delp lphin hinee Mess Messot ot and and Et Etien ienne ne Roesch assisted with the data collection and the transcriptions. tions. This resear research ch was funded funded by the HUMAIN HUMAINE E IST Framework VI Network of Excellence. References Axels Axelsson son,, A.S A.S., ., 20 2002. 02. Di Digit gital al divide divide—st —stat atus us diffe differen rence cess in vir virtua tuall environm envi ronment ents. s. In: Schroede Schroeder, r, R. (Ed. (Ed.), ), The Social Social Life of Avat Avatars: ars: Presence and Interaction in Shared Virtual Environments. Springer, London, pp. 188–204. Bargh, J.A., McKenna, K.Y.A., Fitzsimons, G.M., 2002. Can you see the real me? Activation and expression of the ‘‘true self’’ on the Internet. Journal of Social Issues 58, 33–48. Bers, M., Gonzalez-Heydrich, J., DeMaso, D., 2001. Identity construction environments: supporting a virtual therapeutic community of pediatric patients undergoing dialysis. In: Proceedings of the CHI. ACM Press, New York, pp. 380–387. Boehner, K., Hancock, J.T., 2006. Advancing ambiguity. In: Proceedings of the CHI. ACM Press, New York, pp. 103–106. Boren, M.T., Ramey, J., 2000. Thinking aloud: reconciling theory and practice prac tice.. IEEE Transac Transactio tions ns on Professi Professional onal Comm Communic unicati ation on 43, 261–278. Bradley, M.M., Lang, P.J., 1994. Measuring emotion—the self-assessment mannequin mannequi n and the semantic semantic differential. differential. Journal of Behavior Therapy and Experimental Psychiatry 25, 49–59. Bruckman, Bruckma n, A., 1993. Gender Swapping on the Internet Internet.. In: Proceedin Proceedings gs of  the INET. Cheng, L., Farnham, S., Stone, L., 2002. Lessons learned: building and deploying shared virtual environments. In: Schroeder, R. (Ed.), The

common commo n them themes, es, an and d a fram framewo ework rk fo forr res resear earch. ch. In: Sch Schro roede eder, r, R. (Ed.), (Ed.), The Social Life of Avatars Avatars:: Presence Presence and Interact Interaction ion in Shared Virtual Environments. Springer, London, pp. 1–18. Sundstrom, P., Stahl, A., Hook, K., 2007. In situ informants exploring an em emoti otiona onall mo mobil bilee me messa ssagi ging ng syste system m in their their eve everyd ryday ay pr pract actic ice. e. International Internat ional Journal of Human– Human–Compute Computerr Studies 65, 388–403 388–403.. Taylor, Tay lor, T.L. T.L.,, 2002. 2002. Livi Living ng digi digital tally: ly: embo embodim diment ent in virtual virtual worlds. worlds. In: Schro Schroede eder, r, R. (E (Ed.) d.),, The Social Social Life Life of Avata Avatars: rs: Presen Presence ce and Interaction in Shared Virtual Environments. Springer, London, pp. 40–62. Taylor, Tay lor, T.L., T.L., 2003. 2003. Inte Intentio ntional nal bodies: bodies: virtual virtual envi environm ronment entss and the designer desi gnerss who shape shape them. them. Internat Internationa ionall Journal Journal of Engineer Engineering ing Education 19, 25–34. Tracy, Tracy, J.L. J.L.,, Robins, Robins, R.W., R.W., 2004. 2004. Putt Putting ing the self into self self-con -conscio scious us emotions: a theoretical model. Psychological Inquiry 15, 103–125. Turkle, S., 1995. Life on the Screen: Identity in the Age of the Internet. Simon & Schuster, New York. Vasalou Vasa lou,, A., Joinson, Joinson, A., Pitt, Pitt, J., 2007. 2007. Cons Construc tructin ting g my onli online ne self self:: avatars avat ars that that incr increase ease self self-foc -focuse used d attenti attention. on. In: Proceedi Proceedings ngs of the CHI. ACM Press, New York, NY, pp. 445–448. Yee, N., Bailenson, J., 2007. The proteus effect: the effect of transformed self-representat self-rep resentation ion on behavior. behavior. Human Communication Communication Research 33, 271–290.