topical media & game development
research directions -- engaging with fictional characters
What do you need to evaluate your game or multimedia application?
There are many ways to gain insight in how your system is being used,
see
section 10.3.
But if you want to establish functional properties of a multimedia application,
for example the effectiveness of using an agent in navigating a virtual environment,
in a scientifically more rigorous way, you need to have:
experimental validation
- a theory -- in our case: PEFiC
- a test scenario -- for example, memory tasks in a digital dossier
- the technology -- to realize applications
In this section, I will briefly describe our efforts in experimentally
validating the use of ECAs in virtual environments.
As technology, we use our
intelligent multimedia technolgy,
described in
sectiona 8.3 and
appendix E.
So what must be explained is the theory we adopt and the test scenarios we use.
PEFiC is a theory developed by Johan Hoorn and Elly Konijn,
to explain
Perceiving and Experiencing Fictional Characters, see
[PEFIC].
The PEFiC theory
may serve as the basis for the experimental evaluation of user responses to embodied agents.
In summary, PEFiC distinguishes between three phases, encoding, comparison and response,
in analyzing the user's behaviour towards an agent.
Encoding involves positioning the agent (or fictitious character)
on the dimensions of ethics (good vs bad),
aesthetics (beauty vs ugliness) and epistemics (realistic vs unrealistic).
Comparison entails establishing personal relevance and valence towards the agent.
Response, finally, determines the tendency to approach or avoid the character,
in other words involvement versus distance.
In general, having a virtual environment, there is, for developing test
scenarios, a choice between:
validation scenario(s)
- navigation -- pure interactivity
- guided tours -- using some narrative structure
- agent-mediated -- navigation and guided tours
For our application, a virtual environment
of an artist's atelier,
we have three experimental conditions,
navigation without an agent, navigation with a realistic agent
and navigation with a cartoon-like (unrealistic) agent.
To ensure that these conditions can be compared,
the acrtual information encountered when using the application
is in all conditions the same.
The independent variable in our experiment, the degree of realism of the agent,
corresponds with the epistemic and to some extent the aesthetic
dimension of appraisal in the PEFiC theory.
As dependent variables we have, among others,
user satisfaction, believability, that is estimated usefulness of the agent,
and also the extent to which the relevant information is retained.
1
The application is a digital dossier for the Dutch artist
Marinus Boezem.
The spatial metaphor we used for the dossier is the artist's atelier.
We created a virtual environment containing a display of the artworks, in 3D,
a file cabinet with textual information,
a workbench for inspecting the artist's material,
and a video projector, with which the user can display a video-recorded
interview with the artist.
The actual task to be performed by the user is to learn
what constraints do apply to the installation of one
of the artworks, Stone and Feather:
Stone and Feather
- feather: 70 cm, from ostrich, curved
- stone: 13.5 cm, white marble
- position: alignment with pedestal, no glue
- environment: 50 lux of light max.
The items mentioned in this list must be reproduced by
the user in a subsequent memory test,
and in another experiment the user must be able
to choose the right materials and reconstruct the artwork.
Our assumption in designing this test scenario
was that the gestural nature of positioning the artwork
will be favorable for the condition with a gesturing agent,
whereas believability will be positively affected
by the degree of realism of the agent.
(C) Æliens
04/09/2009
You may not copy or print any of this material without explicit permission of the author or the publisher.
In case of other copyright issues, contact the author.
