Triple-View: Improving Persuasion in Group Video Conferencing through spatial Faithfulness



Similar documents
Eye-contact in Multipoint Videoconferencing

Visual Filler: Facilitating Smooth Turn-Taking in Video Conferencing with Transmission Delay

How To Use Eye Tracking With A Dual Eye Tracking System In A Collaborative Collaborative Eye Tracking (Duet)

A QoE study of different stream and layout configurations in video conferencing under limited network conditions

2003 Midwest Research to Practice Conference in Adult, Continuing, and Community Education

Eye contact over video Kjeldskov, Jesper; Skov, Mikael; Smedegaard, Jacob Haubach; Paay, Jeni; Nielsen, Thomas S.

VACA: A Tool for Qualitative Video Analysis

The Social Dilemma Game - A qualitative Analysis

Reading with Mobile Phone & Large Display

AR-based video-mediated communication:

Marketing oneself: what do small business owners look for when interviewing job candidates?

RESEARCH ON SPOKEN LANGUAGE PROCESSING Progress Report No. 29 (2008) Indiana University

Analysing Questionnaires using Minitab (for SPSS queries contact -)

Figure 1. An embedded chart on a worksheet.

VIRTUE The step towards immersive telepresence in virtual video-conference systems

Least Squares Estimation

Video Conferencing Display System Sizing and Location

Identification of Critical Success Factors for Successful TQM Implementation in Textile Industries, Pakistan

Interaction Techniques for Co-located Collaborative TV

1 Introduction. Hyewon Lee 1.1, Jung Ju Choi 1.2, and Sonya S. Kwak 1.3,

Eye Contact in Leisure Video Conferencing. Annick Van der Hoest & Dr. Simon McCallum Gjøvik University College, Norway.

One-Way Pseudo Transparent Display

Computer Mediated Communication (CMC) Group members use of computers in order to communicate with one another

Towards Inferring Web Page Relevance An Eye-Tracking Study

GAZETRACKERrM: SOFTWARE DESIGNED TO FACILITATE EYE MOVEMENT ANALYSIS

An Investigation on Learning of College Students and the Current Application Situation of the Web-based Courses

How To Find Out If You Can Be Successful In A Career In Physical Education

RECRUITERS PRIORITIES IN PLACING MBA FRESHER: AN EMPIRICAL ANALYSIS

The Power of Dots: Using Nonverbal Compensators in Chat Reference

Presentation of Visual Art in Interactive 3D Environments

NORTHERN VIRGINIA COMMUNITY COLLEGE PSYCHOLOGY RESEARCH METHODOLOGY FOR THE BEHAVIORAL SCIENCES Dr. Rosalyn M.

Data Analysis Tools. Tools for Summarizing Data

Using SPSS, Chapter 2: Descriptive Statistics

Meeting Mediator: Enhancing Group Collaboration with Sociometric Feedback

Video-Based Eye Tracking

Projects Involving Statistics (& SPSS)

STUDENT PERCEPTIONS OF INSTRUCTOR INTERACTION IN THE ONLINE ENVIRONMENT

NETWORK REQUIREMENTS FOR HIGH-SPEED REAL-TIME MULTIMEDIA DATA STREAMS

KADI SARVA VISHWA VIDYALAYA GANDHINAGAR. Ph.D. Course Work SOCIAL WORK

STUDENTS PERCEPTIONS OF TEACHER CREDIBILITY AND LEARNING EXPECTATIONS IN CLASSROOM COURSES WITH WEBSITES

Social Presence Online: Networking Learners at a Distance

Conference interpreting with information and communication technologies experiences from the European Commission DG Interpretation

Faculty Guide to Teaching through Videoconferencing. Prepared by Allan Gyorke Education Technology Services, Penn State University

SYNCHRONOUS COMMUNICATION AND IMMEDIACY 1. Synchronous Communication and Immediacy in the Online Classroom: A Call for Research and Practice

Kevin Mawhinney, Technology Education Department Head, Cobequid Educational Centre, Truro, Nova Scotia,

Videoconferencing Design for Remote Groups

MirrorSpace: using proximity as an interface to video-mediated communication

3. INFORMATION SYSTEMS INFRASTRUCTURE AND EQUIPMENT

The Online University Business School - Athabasca University s Centre For Innovative Management

Effects of Orientation Disparity Between Haptic and Graphic Displays of Objects in Virtual Environments

INVOLVING STAKEHOLDERS IN THE SELECTION OF A PROJECT AND PORTFOLIO MANAGEMENT TOOL

When you re far away, your own image

TRACKING DRIVER EYE MOVEMENTS AT PERMISSIVE LEFT-TURNS

Calculating, Interpreting, and Reporting Estimates of Effect Size (Magnitude of an Effect or the Strength of a Relationship)

Briefing document: How to create a Gantt chart using a spreadsheet

An introduction to videoconference teaching

Adaptive Tolerance Algorithm for Distributed Top-K Monitoring with Bandwidth Constraints

Using Excel for inferential statistics

Directions for using SPSS

MARKETING EDUCATION: ONLINE VS TRADITIONAL

Data Coding and Entry Lessons Learned

Transcription:

Triple-View: Improving Persuasion in Group Video Conferencing through spatial Faithfulness Apoorva Sachdev University of California, Berkeley Berkeley, CA apoorvas@berkeley.edu John Canny University of California, Berkeley Berkeley, CA jfc@cs.berkeley.edu Abstract Video conferencing is integral to most businesses these days as teams are spread all around the world. However, video conferencing can negatively affect high stakes communications. Prior work suggests that spatial distortions of non-verbal cues, particularly gaze and deixes, negatively impact many aspects of effective communication in dyadic situations. In this study, we examine the effects that spatial distortion of nonverbal cues has on persuasive power. We conducted the study and found that use of traditional video-conferencing system, that introduce spatial distortions negatively affect persuasive power of participants during communication. On the other hand, use of Triple-View, a spatially faithful video conferencing system, eliminates such effects and enables participants to be more persuasive. Copyright is held by the author/owner(s). Keywords CSCW, CMC, Persuasion, Video Conferencing Spatial Faithfulness, Gaze Awareness, Eye Contact

Figure 2: A conventional-video conferencing set-up for group to-group communication. It uses one screen display and one camera. The dashes lines indicate the perceived positions of remote participants. ACM Classification Keywords H.5.3 [Information Interfaces and Presentation]: Group and Organization Interfaces Computer-Supported Cooperative Work; H.4.3 [Information Systems Applications]: Communications Applications Computer Conferencing, Teleconferencing, and Videoconferencing General Terms Design, Experimentation, Human factors Introduction Live face-to-face interaction is argued to be the most powerful medium for persuasion. However, as world is becoming more global and technology-reliant, we are becoming more and more dependent on computermediated communication. Especially in a business setting, where remote communication between teams is integral, having an effective communication interface is fundamental. Video-conferencing systems, in particular are widely used to facilitate remote communication. However, traditional videoconferencing systems do a poor job of preserving nonverbal cues that are important in group communication. Gaze and deictic gestures, like many non-verbal cues, are dependent on spatial faithfulness of the video system. Spatial faithfulness refers to the extent to which a system preserves spatial relationships and allows participants to maintain eye-contact. Hence, a spatial faithful video-conferencing system preserves spatial faithfulness and can effectively send across directionality-dependent data. It has been shown that maintaining eye-contact/gaze increases credibility and thereby would greatly affect persuasiveness [14]. We extend those finding and test them in the videoconferencing scenario to show that a spatially faithful system leads to increased persuasive power when compared to a conventional video-conferencing system. Traditional video conferencing systems use a singleshared display, so all people on one end see the same view independent of their perspective. This is known as perspective invariance and its cognitive mechanisms are well-understood [17]. Consider a scenario shown in Figure 2, where a group-to-group meeting is taking place. The dotted characters represent the images of the remote participants on the screen. Suppose Participant L gazes at Participant 2 on the screen. Since the perceived position of Participant 2 is right under the Camera 2, Participant 2 will register correctly that Participant L is looking at him/her. However, so will Participants 1 and 3, since they take on the shared perspective of Camera 2 independent of their actual viewing angle. Hence, each participant on the remote end will simultaneously register direct eye contact with Participant L. The current system, Triple-View tries to prevent this from happening. It is built upon the design of Multi-view [6], a spatially faithful system. Using this display, the users can correctly identify gaze and gesture direction across the video boundary. In this paper, we consider the affect spatial faithfulness has on the persuasive power of participants in remote group communications by comparing conventional videoconferencing system and the Triple-view videoconferencing system.

3 Figure 4: Three remote participants are gazing at viewing position 1 (see Figure 3). Column 1 is the view from position 1, column 2, position 2, and column 3, position 3.. The top row shows the Triple-View display with appropriately changing perspectives. The bottom row shows what is seen from the respective positions with non-directional video conferencing and demonstrates perspective invariance Triple View Conventional Figure 3: A Triple-view video - conferencing set-up for group to-group communication. It uses 3-cameras to capture three unique perspectives, which correspond to the correct perspectives of the remote participants. It uses a 3-way display to allow remote participants to view their respective perspectives simultaneously. Triple-View Design In face-to-face communication, each participant has their own unique perspective defined by their position. Triple-view tries to replicate this in the group to-group (3-3) communication case and prevent perspective invariance. It uses a 3-way viewpoint directional display to simultaneously display different video streams to different participants based on their viewing position. Multiple cameras are used to capture unique perspectives for each participant and then provide fullspatial faithfulness for all participants. Consider the same scenario as before but now using a Triple-view display as shown in Figure 3. The dotted characters represent the images of the remote participants on the screen. By virtue of the Triple-view system, when Participant 3 looks at the screen she sees the video captured by camera 3. Now, when Participant L gazes at Participant 2 on the screen, Participant 2 will register correctly that Participant L is looking at him/her as Participant 2 is viewing the video captured from Camera 2. However, when Participants 1 and 3 look at the display, they see the view through Camera 1 and Camera 3 respectively, and see the Participant L gazing to the right, in the case of Participant 1 and gazing Left, in the case of Participant 2. The different views can clearly be seen in the Figure 4 more clearly as seen by the system. Related Work Video conferencing System Hydra [8] is a similar system that supports multi-party conferencing by providing a camera/display surrogate that occupies the space that would otherwise be occupied by a single remote participant. Because each person is presented by their own screen, it is possible to preserve gaze, though the size of the image and the system itself is small. Gaze-2 [9] supports gazeawareness for multi-user eye-tracking by using an array of cameras and tracking exactly which camera the user is looking directly at. In this case, the remote user s orientation and gaze is represented by rotating the planar frontal views of the other participants. MAJIC [10] is another implementation that reduces the parallax error introduced by positioning cameras around the screen during video-conferencing by placing the camera behind a semi-transparent screen. This

4 Figure 5: The NASA Moon Survival Problem asks people to rank the order of usefulness of these 15 items and then compares the list with the NASA expert solution to determine the quality of the ranked list. allows remote users to effectively track gaze and hold eye-contact. The common limitation for all these approaches is that they are designed to support multi single-participant sites instead of many multiparticipant sites and Tripleview tries to solve this limitation. Eye-View [7] deals with bringing in focus the person who is talking and readjusting the video-size/screen arrangement depending on the number of participants. This allows multiple people to communicate and focus attention on one dominant person, however does not preserve gaze. MultiView [6] is a similar system created by David Nyugen, which uses projectors, multiple cameras and a special reflective screen to preserve the gaze of the participants. This system allows for multi-participant sites but offers poor imagequality not good enough for perception of precise eyecontact and the setup requires precise placement and use of expensive projectors. Triple-view is the third generation iteration of the MultiView that offers much better resolution and hence more spatial fidelity. Since, it is inherently a 3-way screen; it doesn t use projectors and it can be potentially used in areas even with bright light. Persuasion and Importance of Spatial Faithfulness David Nyugen et al. in [11] proves how directional video conferencing can reduce the delay and fragility in trust formation in comparison to non-directional video conferencing by evaluating the performance of teams in social dilemma scenarios. GA display [12] is another study that claims that having full gaze awareness can reduce the number of words required to communicate a task by half. Peter j. Werkhoven in his study evaluated 3 conferencing scenarios, namely non-isotropic videoconferencing, isotropic video-conferencing and face-toface system using Lost on the moon survey. Their implementation of the isotropic video-conferencing system was different from Triple-view as they virtually placed participants in a triangular configuration in order to facilitate the chat [4]. A similar study as in [4] was used to evaluate Triple-view video-conferencing system. Experiment Setup Our evaluation seeks to understand if use of Triple View has an impact on persuasion. One common task used for persuasion studies is the NASA Moon survival problem [2] [3] [4]. Participants have group discussion about a common strategy to survive on the moon and then rank 15 items in order of their usefulness (for survival on the moon). We adapted this problem for our study, by introducing a remote confederate, who video conferences with the group and argues for a particular wrong ordering of items. The study was set up as shown in Figure 5, with one remote confederate and 3 participants. It is an asymmetric set-up due to availability of only one Triple-view screen, thus only the 3-participant location has a triple-view screen. The study was conducted as follows: Figure 6: the experimental setup for 1- group communication. A triple-view screen is used on one end while a normal one-view display is used on the other. Each of the participants was asked to solve the NASA Moon Survival problem individually i.e. rank the 15 items. [5 mins] The 3 participants then collaborated amongst themselves to come to a consensus about the final ranking of the 15 items. [10 mins] They then did a video-call with the confederate (depending on the category either using Triple-view or conventional video conferencing system) and were

Spearman's Rank Correlation Coefficient 5 Figure 7: The user-study setup. Above confederate setup Below participants with Triple-View. 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Before Confederate Interaction Conventio nal After Confederate Interaction Figure 8: The Spearman's rank correlation co-efficient for groups ranking order and NASA expert solution with 95% confidence interval shown in error bars. persuaded to change the rankings as they conversed [10 mins] The participants finalize their rankings after videocall. [5 mins] The study was performed in groups of 3 participants and on a total of 6 groups. The 6 groups were divided as follows: Category A [Control] - use conventional video conferencing system i.e. non-directional. Category B [Experimental] use Triple-view video conferencing system. The confederate remained the same during all the studies. Our hypothesis is that Triple-View conferencing system will allow the confederate to be more persuasive than the conventional video-conferencing system as it preserves gaze, directionality of non-verbal cues and ultimately improves spatial fidelity. The interaction offered would be more natural and realistic and hence allow for better collaboration and negotiation between participants. One measure of persuasion captures the group s movement towards the confederate s ranking by comparing the group s pre-confederate and postconfederate interaction ranking with the expert s ranking. We used the standard rank correlation method, Spearman's rank correlation co-efficient. Spearman's rank correlation co-efficient takes into account the difference between the rankings of the expert and the groups before and after the interaction and measures the movement as follows: M = d(u, E) d(u, E) where E is the expert s ranking and u, u+ are the groups ranking before and after the persuasion event. Post-Session questionnaire After the study, the participants were asked to fill out a questionnaire about their experiences interacting with the system. There were 5 Likert scale questions with responses on a scale 1 (strongly disagree) to 5(strongly disagree) and 5 open-ended questions asking them to describe the interaction. Results Comparison of systems: Table 1: Spearman's rank correlation co-efficient for groups initial and final rankings against the expert s ranking Video System Spearman rank correlation coefficient Before Interaction Spearman rank correlation coefficient After Interaction Mean Variance Mean Variance Conventional 0.678 0.004 0.345 0.004 Triple-view 0.712 0.003 0.532 0.005 The result of a t-test done on pre-interaction Spearman s rank correlation co-efficient is nonsignificant as (p>0.05). This is expected, as the initial rankings of the groups are independent of the videosystem used later in the study. This serves as a

6 Figure 9: The qualitative data from users for both the video-conferencing systems. *only groups who used the Triple-view system were asked this question baseline to examine changes seen after the interaction. As can be seen from the table above, the average Spearman s rank correlation co-efficient value for the group s post-interaction rankings was lower in the Triple-view case than in the conventional videoconferencing system case. A t-test performed on the post-interaction Spearman s rank correlation coefficient shows that the results are significant as p < 0.05. However, the p-value is close to 0.05 (0.037), so the study may need to be performed on a larger sample size to get conclusive results. Post-Session Questionnaire As can be seen in Figure 9, the answers are not drastically different; however, we did find that participants using Triple-view were better able to detect the direction of gaze of the confederate more accurately and were regularly able to maintain eyecontact than people using the conventional videosystem. We found no particular difference in how believable the confederate was in both the conditions and how the participants felt about changing their minds about rankings. Thus, it is valid to consider confederate as a controlled variable for the purpose of this study as we can say that he was able to maintain his style and level of persuasion constant irrespective of the video conferencing system used. Discussion Our hypothesis is supported by the results shown above. The quantitative results we found were significant and hence indicate that Triple-view enabled the confederate to be more persuasive than conventional video conferencing system and are qualitative results also support the hypothesis. For future work, we plan to study how involvement and participation of group members changes depending on the video-conferencing system used. Conclusion In this paper, we described a study comparing the persuasive power of two video-conferencing systems. We presented a new iteration of a previously designed video-conferencing system, which preserves many of the non-verbal cues that are usually lost in standard video conferencing systems. We found that Triple-view, a system that preserves spatial faithfulness enables participants to be more persuasive than the conventional video conferencing systems. Participants that interacted with the confederate through directional video conferencing seem to change their rankings in accordance to the suggestions made by the confederate more often in comparison to nondirectional as shown in our results. Acknowledgements We thank all the anonymous reviewers and participants for their feedback and all the BiD lab members for their support and co-operation. We also thank Professor Hartmann and Drew Fisher for their continuous guidance and support. References [1] Vishwanath, D., Girshick, A.R., and Banks, M.S. Why pictures look right when viewed from the wrong place. Nature Neuroscience 8, 10 (2005), 1401 1410 [2] Hall, J., Watson, W.H.: The Effects of a Normative Intervention on Group Decision Making Performance. Human Relations 23, 299 (1970) [3] Bradner, E., Mark, G.: Why distance matters: Effects of cooperation, persuasion and deception. In:

7 Proceedings of Computer-Supported Collaborative Work 2002, pp. 226 235. ACM Press, New York (2002) [4] Werkhoven, P., Schraagen, J.M., Punte, P.A.J.: Seeing is believing: communication performance under isotropic teleconferencing conditions. Displays 22 (2001) [5] Ramachandran, D. and Canny, J. The persuasive power of human-machine dialogue. In H. O.-K. et al., editor, PERSUASIVE 2008, LNCS 5033, pages 189 200, 2008 [6] David Nguyen and John Canny. MultiView: Spatially Faithful Group Video Conferencing. Proceedings of the 2005 Conference on Human Factors in Computing Systems (CHI 05) pp. 512-521. Portland, Oregon. [7] Tracy Jenkin, Jesse McGeachie, David Fono and Roel Vertegaal, eyeview: Focus+Context Views for Large Group Video Conferences, conference on Human factors in computing systems, 2005, New York. [8] Sellen, A., Buxton, B., and Arnott, J. Using spatial cues to improve videoconferencing. Proc. CHI 1992, ACM Press (1992), 651-652. [9] Vertegaal, R., Weevers, I., Sohn, C., and Cheung, C. Gaze-2: conveying eye contact in group video conferencing using eye-controlled camera direction. Proc. CHI 2003, ACM Press (2003), 521-528, New York. [10] Okada, K., Maeda, F., Ichikawaa, Y., and Matsushita, Y. Multiparty videoconferencing at virtual social distance: MAJIC design. Proc. CSCW 1994, ACM Press (1994), 385 393. [11] David Nguyen, and Canny, J. MultiView: Improving Trust in Group Video Conferencing through Spatial Faithfulness. Proceedings of the 2007 Conference on Human Factors in Computing Systems (CHI 07). San Jose, CA. [12] Monk, A., and Gale, C. A look is worth a thousand words: Full gaze awareness in video-mediated conversation. Discourse Processes 33, 3 (2002), 257-278. [17] Burgoon, J. K., Birk, T. and PFAU, M. (1990), Nonverbal Behaviors, Persuasion, and Credibility. Human Communication Research, 17: 140 169