Engaging Virtual Pedagogical Agents (VPAs) Agneta Gulz Div. of Cognitive Science (LUCS) Lund University agneta.gulz@lucs.lu.se Magnus Haake Dept. of Design Sciences, LTH Lund University magnus.haake@design.lth.se
Adele [ Teacher / Coach ]
Cosmo [ Instructor ]
EduAgent [ Learning companions ]
Steve [ Instructor ]
Steve [ Team member / learning companion ]
Sam [ Learning companion ]
Characters from FearNot [ Learning companions ]
Laura the Exercise Advisor [ Instructor / Coach ]
[ Virtual speech therapists ]
Anna (IKEA) [ Virtual assistant / Webbot ]
Sara (Malmö) [ Virtual City guide / Webbot ]
Hanna (Försäkringskassan) [ Virtual assistant / Webbot ]
The Microsoft Office Assistant (Clippit)
A central goal of VPAs is to increase learners engagement.
Indeed there is intuitive appeal in the argument that if a learning environment or material is found engaging that is, experienced as involving, interesting or as having impact it is likely that users will become more active, stay on longer and produce more, which may, in turn, have effects on learning achievements. Citation: Gulz, A. & Haake, M. (2006). Design of animated pedagogical agents a look at their look. Int. J. of Human-Computer Studies, 64(4).
Engagingness in this sense is not to be equated with entertainment. To be entertained by an interface agent does, indeed, imply an important form of engagement, but one can be engaged on several other grounds as well. The impact that an engaging program has does not even have to be of a pleasant nature. Citation: Gulz, A. & Haake, M. (2006). Design of animated pedagogical agents a look at their look. Int. J. of Human-Computer Studies, 64(4).
Learner engagement being a central goal, it is surprising how relatively little the (static) visual aspects of VPAs have been researched. Static visual aspects: e.g. graphical style, body and face shape, gender, ethnicity, clothes & attributes, colours & textures. Dynamic visual aspects: e.g. gestures, facial expressions, speech, gaze. Reference: Gulz, A. & Haake, M. (2006). Design of animated pedagogical agents a look at their look. Int. J. of Human-Computer Studies, 64(4).
Degree of visual naturalism Degree of Humanness Graphical Style: Naturalism Stylization Basic Physical Properties Graphical Style: Detailed Simplified
The topic for this presentation Degree of naturalism in the visual design of VPAs in relation to engagability
Arguments for visual naturalism The smooth communication argument The identification argument The believability argument Arguments against visual naturalism The false expectations argument The problem of representing perfection
Increased visual naturalism as a means to achieve increased engagement in users There is a desire to attain more and more [visually] realistic virtual worlds and characters in computer games [ ] One reason which is pointed out very often is the belief that the reduction of the absolute difference between real and virtual environments leads to an increase of presence and of immersion. Citation: Wages et al. (2004). How realistic is realism? Considerations on the aesthetics of computer games. In Proc. ICEC 2004.
Increased visual naturalism as a means to achieve increased engagement in users (continued) Pictorial realism increases involvement and the sense of presence in a digital environment. Pictorial realism may even be a condition for human cooperation with an animated/virtual agent. Reference: Welch et al. (1996). The effects of pictorial realism, delay of visual feed-back, and observer interactivity on the subjective sense of presence. Presence, 5.
Increased visual naturalism as a means to achieve increased engagement in users (continued) More visually realistic characters-agents may increase the users curiosity about the personality of the characters. Reference: Wilson, M. (1997). Metaphor to personality: the role of animation in intelligent interface agents. Presented at IJCAI-97.
Increased visual stylization as a means to achieve increased engagement in users. People more easily get involved with, elaborate on, and even project themselves into a character that is not visually naturalistic. A highly naturalistic character is a visual and socio-emotional fact, which does not leave much for the user to fill in. A stylized character, on the other hand, invites elaboration by the user, who may fill in from his/her own personal and subjective experience. Reference: McCloud, S. (1993). Understanding Comics: The Invisible Art. Harper Perennial, NY, USA.
Increased visual stylization as a means to achieve increased engagement in users. (continued) The stylized character is, [ ] an empty shell that we inhabit. Citation: McCloud, S. (1993). Understanding Comics: The Invisible Art. Harper Perennial, NY, USA.
Learners were allowed to make a choice between: (i) more naturalistic vs. more stylized virtual pedagogical agents, as well as between (ii) strictly task oriented vs. more socially oriented (taskand-relation oriented) virtual pedagogical agents. There was a significant correlation between preferences for socially oriented and visually stylized agents. The theory of McCloud served as a possible framing that could help explain or interpret this result. Reference: Gulz, A. & Haake, M. (2005). Social and Visual Style in Virtual Pedagogical Agents. In Workshop Proc., UM 05.
Program dummy: Open College Sweden: Education Online
Participant utterances on choices and on the two visual styles I prefer these [stylized ones]; there is more left for your own imagination here you can express your personality or whatever you have inside. Well, these, the Manga ones, are more for you heart, because they really concern personality. With these [stylized ones] you can be whoever you like and you don t have to reveal yourself to people that you don t know.
Furthermore 6 out of the 21 participants 1 choosing a naturalistic and 5 choosing a stylised character indicated that they with their choice of character made use of the possibility to look different and/or be different.
An interpretation of this result is that a visually stylized character leaves more for peoples imagination and their projections of a fantasy self into the character.
Proposal Visually stylized characters increase users/learners active participation in terms of intellectual as well as socioemotional engagement.
However, even if this relation turns out to be general and stable, our conclusion would not be that one should go for stylized characters only!
Magnus Haake Dept. of Design Sciences, LTH Lund University magnus.haake@design.lth.se Agneta Gulz Div. of Cognitive Science (LUCS) Lund University agneta.gulz@lucs.lu.se
References Gulz, A., and Haake, M. Design of animated pedagogical agents a look at their look. Int. J. of Human-Computer Studies, 64, 6 (2006), 322-339. McCloud, S. (1993). Understanding Comics: The Invisible Art. Harper Perennial, NY, USA. Wages, R., Grunvogel. S. M., and Grutzmacher, B. How realistic is realism? Considerations on the aesthetics of computer games. Proc. ICEC 2004, Lecture Notes in Computer Science, 3166/2004, Springer- Verlag (2004), 216 225. Welch, R., Blackmon, T., Liu, A., Mellers, B., Stark, L., 1996. The effects of pictorial realism, delay of visual feedback, and observer interactivity on the subjective sense of presence. Presence: Teleoperators and Virtual Environments, 5, 263-273. Wilson, M. (1997). Metaphor to personality: the role of animation in intelligent interface agents. Presented at Animated Interface Agents: Making them Intelligent (in conjunction with IJCAI-97), Nagoya, Japan.