Page 0: Page 1: Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone Sin-Hwa Kang, James H. Watt, and Sasi Kanth Ala 윤 혁기 1 / 27 Page 2: Abstract Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  This paper research the effect of behavioral and visual realism of avatars on users’ social copresence in emotionally engaged conversations conducted via a simulated mobile video telephone.  Social Copresence to better measure user’s engagement with conversational partners in mobile telephone & anonimity & nonverbal environment  Static / Dynamic, High / Low iconic 2 / 27 Page 3: Overview Social Bond Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone Social Connectedness Presence Social Presence Social Copresence Avatar Anonimity Copresence Mobile Communication Nonverbal Communication 3 / 27 Page 4: Introduction  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Social Connection and Mobile Communication Humans’ social bonds are elicited by their basic need to be connected with others, which can be called social connectedness. – Smith and Mickie Social connectedness is defined as the presence of relationships among people which are essential to their health and successful life and are fostered when those relationships are maintained constructively . - Ministry of Social Development in New Zealand Mobile communication has allowed people to make dynamic plans and construct fluid social networks through functional and expressive management – Ling   4 / 27 Page 5: Introduction  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Social Connection and Mobile Communication(Cont.) This sense of being co-connected is likely to be affected by the addition of a visual element to voice communication In this research, we will investigate methods of preserving visual anonymity while retaining nonverbal information.  5 / 27 Page 6: Introduction Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Importance of Nonverbal Communication in Mediated Interactions  Nonverbal behavior is used as a primary channel to communicate emotion and to indicate changes in “the quality of an ongoing interpersonal relationship.” – Ekman Friesen Many researches shows, the facial cues convey emotional signals most efficiently – Littlejohn, Izard, Poggi, Pelachaud... But whether presenting such cues on the small screen of a mobile communication device is effective remains unanswered   6 / 27 Page 7: Introduction  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Presence, Copresence, and Social Copresence Effective mediated interaction can be obtained by “the creation of social presence and the improvement of social communication.” – Biocca Harms Social presence is..  “Salience(양상) of others in mediated communication and consequent salience of their interpersonal interaction”  “Interaction in mediated environments”  “one’s model of the other intelligence”   “access to another intelligence”  Copresence is..  “the concept which describes people’s engagement in being and doing together in such as Collaborative Virtual Enviroment”  “co-situated within the same interpersonal enviroment.” 7 / 27 Page 8: Introduction  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Presence, Copresence, and Social Copresence (Cont.) In this view, Social Copresence can be conceived of as involvement and engagement through mutual awareness between intelligent beings who have a sense of access to the other being consciously, psychologically and emotionally, within a mediated environment perceived as capable of supporting social communication. Three elements of Social Copresence  Copresence – users’ sense of being connected with their partner  Social Richness of Medium(매개물) – perceived capability of the medium to connect interactants socially  Interactant Satisfaction – presence of social attraction and emotional credibility between interactants  8 / 27 Page 9: Introduction  Anonymity and Avatars  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone Avatars is “digital models of people that either look or behave like the people they represent” - Bailenson Avatar’s two function is..  Represent the communicator and to communicate some information about him/her  To mask the visual identity of the communicator, unless a photographic image or video of the communicator is used as an avatar   The use of avatars promotes presence among users in some mediated environments. 9 / 27 Page 10: Introduction  Anonymity and Avatars  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone Conditions of complete anonymity provide freedom from routinized social ties - Milgram Avatars’ nonverbal behavior provides a channel for transmitting the emotional signal - Garau Communication effectiveness is increased when people look at facial expressions during their interactions. – Poggi Pelachaud, Walker, Sproull, Subramani Avatar’s realism is generally defined as possession of either the visual appearance or behavioral attributes of a human – Nowak, Rauh    10 / 27 Page 11: Research Hypotheses  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  The study reported here differs from pre-study.. In investigating the impact of different avatar presentations on a small screen typical of mobile communication devices In investigating the impact of avatars in a communication task that has an affective rather than an instrumental goal   Static avatars are less behaviorally realistic than dynamic ones.  Dynamic avatars that convey facial expression would be expected to produce more social copresence than those that don’t 11 / 27 Page 12: Research Hypotheses Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  H1 : People will report higher Social Copresence when interaction partners are represented by more behaviorally realistic dynamic avatars, rather than by static avatars.  H2 : Interactants’ Social Copresence will be higher when interaction partners are represented by nonanonymous avatars, rather than by anonymous avatars.  H3 : Interactants’ Social Copresence will be higher when interaction partners are represented by highiconic anonymous avatars, rather than by low-iconic anonymous avatars. 12 / 27 Page 13: Experimental Design  Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  2 x 5 x 3 factorial between-subject experiment Behavior realism  Dynamic Avatar  Static Avatar  Visual Realism or avatar type  Unmodified Video  Overall Processed Video  Outer Masked Processed Video  High Humanoid Avatar  Low Humanoid Avatar Visual Realism or avatar type  Gender dyad type  Male-Male / Female – Female / Male - Female 13 / 27 Page 14: Experimental Design Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  2 x 5 x 3 factorial between-subject experiment 14 / 27 Page 15: Experimental Design  Normal arm length distance Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Image Displayed on laptop computer simulated mobile telephone with a realistic small-size screen  Normal view conditions on actual video mobile telephone  Hands-free mobile telephone headset, providing user for the audio 15 / 27 Page 16: Experimental Design Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Overall Processed Video experimental condition was produced by implementing a real-time edge detector  Edge-detected video produced lower identification of video-recorded actors than the video pixelization that is often used to hide identities in new broadcasts – Zhao and Stasko  In sample pilot test, 3 / 17 (18%) correctly identified the person. 16 / 27 Page 17: Experimental Design Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  In Outer Masked Processed Video experimental condition, Peripheral head cues, rather than central facial features, were most important in recognizing faces at low visual resolution – Jarudi and Sinha  To give a condition of maximum anonymity edgedetected video was additionally masked to eliminate periphral head cues  Logitech Video Effect program used to provide high- or low-iconic avatar condition, which generates real-time animation of a graphical avatar’s facial expressions. 17 / 27 Page 18: Experimental Participants and Procedures  One hundred eighty college students were recruited  Paired with someone whom they unlikely to know beforehand  Experimental interaction took place in two separate rooms where the participants were placed separately and at different time, to eliminate initial face-to-face interactions.  Participants were given a hypothetical conversational scenario where they were instructed to assume the role of a student who seeks to find out if the other person is suitable match with whom to shared an apartment and possibly establish a friendly relationship 18 / 27 Page 19: Measurement Instruments  Copresence  Seventeen item Likert-type scale was created with 8 point metric for item (1 = Very Little; 8 = Very Much) This measurement included two separate sets of items Nowak and Biocca  Perceived other’s copresence  Self-reported copresence  “I want to make the conversation more intimate”   Social Richness of Medium  Five item Likert-type scale was created with 8 point metric for item (1 = Very Little; 8 = Very Much) Subjective social richness of the medium - Nowak and Biocca  “To what extent did you feel you could get to know someone that you met only through mobile phone?” 19 / 27  Page 20: Measurement Instruments  Interactant Satisfaction  Fifteen item Likert-type scale was created with 8 point metric for item (1 = Very Little; 8 = Very Much) Six of the items were adpoted from the items of social attaction scale used – in Nowak’s study  “I would like to have a friendly chat with her/him”   Nine item, called emotional credibility(신뢰도), were newly created to measure interactants’ emotional perceptions of their interaction partners – in Kang, Wat, Ala, Isbister’s study  “S/he expresses feelings and emotions appropriately for the situation” 20 / 27 Page 21: Results  Results Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone H1 : People will repo H2 : Interactants' Soc H3 : Interactants' Soci rt higher Social Copr ial Copresence will b al Copresence will be esence when interact e higher when intera higher when interactio ion partners are repr ction partners are re n partners are represe esented by more be presented by non-an nted by high-iconic an haviorally realistic dy onymous avatars, rat onymous avatars, rath namic avatars, rather her than by anonym er than by low-iconic than by static avatars ous avatars. anonymous avatars 1. Copresence 2. Richness of Medium H1.1 : not supported H2.1 : not supported H3.1 : supported H1.2 : supported H2.2 : not supported H3.2 : supported 3. Interactant Satisfaction H1.3 : not supported H2.3 : not supported H3.3 : not supported 21 / 27 Page 22: Results Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  H1.1 : Low-Iconic, Low-Humanoid Avatars produced significantly less Copresence between communicators  H1.1 : This indicates that participants were more satisfied with the conversation when interacting with the same gender partners when the effects of behavioral realism and visual appearance of the avatar were controlled.  H1.2 : A Medium presenting more realistic avatars tended to be seen as more capable of supporting social interaction, with abstract drawings being rated lower. 22 / 27 Page 23: Results Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  H2.2 : Richness of Medium in anonymous and nonanonymous avatars, dynamic avatars were rated significantly higher than static avatars  H3.1 : High-Iconic avatars produce significantly higher copresence than did Low-Iconic avatars  H3.1 : Higher Copresence was reported in samegender dyads.  H3.2 : In high- and low-iconic avatars, Dynamic avatars were judged significantly higher on Social Richness of Medium than were static avatars 23 / 27 Page 24: Conclusion and Discussion  Dynamic avatars were judged more capable in Social Richness of the Medium as were High-Iconic avatars  High-Iconic avatars did produce a higher Copresence evaluation than did Low-Iconic avatars  Avatar Differences, be they behavioral realism, anonymity or iconicity, showed no impact on participants’ Interaction Satisfaction  The result of this study illustrate the incomplete nature of that evaluation process and make the case for including copresence and interaction satisfaction measures when evaluating the full Social Copresence potential of new technologies. 24 / 27 Page 25: Conclusion and Discussion  No difference seen between anonymous and nonanonymous avatars on any of the three elements of Social Copresence  Edge-detection filtering used in this experiment may be the best way to provide both anonymity and effective communication in mobile video telephone. 25 / 27 Page 26: Future Directions Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  Developing such a taxonomy(분류), in which the relevant(적절한) characteristics of each kind of communication task or setting are fully explicated should be a high priority in future research  Identifying the use and impact of visual information in the two contrasting communication settings – task oriented communication and social information, cohesion is also a future research priority 26 / 27 Page 27: Limitations Social Copresence in Anonymous Social Interactions Using a Mobile Video Telephone  This study examined communication between strangers  It also involve role playing, and this may not produce responses that are identical to those produced in actual situations.  Copresence was being evaluated, may have been a function of the small visual display’s limited abilities to convey nonverbal cues  Using a simulation instead of an actual device means that the user’s sensory experience was not identical to what the real device would deliver.  The experimental task was chosen to be relevant to college students 27 / 27 Page 28: