Playing In or Out of Character: User Role Differences
in the Experience of Interactive Storytelling
Christian Roth, M.Sc.,
Ivar Vermeulen, Ph.D.,
Peter Vorderer, Ph.D.,
Christoph Klimmt, Ph.D.,
David Pizzi, Ph.D.,
Jean-Luc Lugrin, Ph.D.,
and Marc Cavazza, Ph.D.
Interactive storytelling (IS) is a promising new entertainment technology synthesizing preauthored narrative
with dynamic user interaction. Existing IS prototypes employ different modes to involve users in a story,
ranging from individual avatar control to comprehensive control over the virtual environment. The current
experiment tested whether different player modes (exerting local vs. global inﬂuence) yield different user ex-
periences (e.g., senses of immersion vs. control). A within-subject design involved 34 participants playing the
cinematic IS drama ‘‘Emo Emma’’
both in the local (actor) and in global (ghost) mode. The latter mode allowed
free movement in the virtual environment and hidden inﬂuence on characters, objects, and story development.
As expected, control-related experiential qualities (effectance, autonomy, ﬂow, and pride) were more intense for
players in the global (ghost) mode. Immersion-related experiences did not differ over modes. Additionally, men
preferred the sense of command facilitated by the ghost mode, whereas women preferred the sense of in-
volvement facilitated by the actor mode.
Computer-based entertainment technologies are re-
markably popular all over the world. While certain video
games such as triple-A shooter or role-playing games repre-
sent the state of the art in computer-based entertainment
technology and dominate markets worldwide, a great
breadth of alternative concepts is evolving as well, such as
browser games, casual games, or mobile/pervasive games.
One particularly promising ﬁeld of innovation in computer-
based entertainment is interactive storytelling (IS). IS systems
present evolving narratives that can be inﬂuenced, in real-
time, by the user.
The basic idea of IS is that intelligent
software dynamically synthesizes preauthored story elements
(e.g., involving characters and their mutual relationships,
events, and locations) with individual user input, allowing
users to cocreate a unique story. By allowing users to exert a
global inﬂuence on a developing storyline, IS is distinct from
conventional video games. Although video games usually
allow users to manipulate local events, character behaviors,
and virtual environments, the storyline itself remains linear,
and often ﬁxed. IS’ vision, in contrast, is more similar to Star
where users generate novel, one-of-a-kind
entertainment experiences that combine characteristics of
advanced video games, virtual reality, and/or (virtualized)
Although the vision of IS is appealing, its development is
still in its infancy. Only few IS prototypes exist, and many
possibilities to design entertaining experiences need yet be
The current research zooms in on a design choice
that may be particularly relevant to IS environments, namely
how to involve players psychologically in the story devel-
opment. In most video games, players are assigned the role of
a protagonist: a local agent focusing on his/her own behavior
and immediate surroundings. In IS, however, perhaps, a
more natural role would be that of a global inﬂuencer: a
writer, director, or mediator whose main goal is to make the
developing story worthwhile. Such a global role could po-
tentially optimize control-related entertainment experiences,
such as autonomy, effectance, and ﬂow.
On the other hand, it
could hamper the more immersive entertainment experiences
that go hand in hand with playing a protagonist: presence,
character identiﬁcation, and affect.
The present study compares experiences of IS users in
a local versus global mode. Employing a previously devel-
of user experience measures derived from
Center for Advanced Media Research Amsterdam (CAMeRA), VU University Amsterdam, Amsterdam, The Netherlands
Department of Communication Science, VU University Amsterdam, Amsterdam, The Netherlands.
University of Mannheim, Mannheim, Germany.
Department of Journalism and Communication Research, Hannover University of Music, Drama, and Media, Hannover, Germany.
School of Computing, Teesside University, England.
CYBERPSYCHOLOGY,BEHAVIOR,AND SOCIAL NETWORKING
Volume 15, Number 11, 2012
ªMary Ann Liebert, Inc.
it aims to discover whether different
ways of involving users in interactive narratives cause sys-
tematically different experiences.
User Roles in Interactive Stories: Actor
versus Ghost mode
Thus far, IS developers emulated common practice in vi-
deo games by arranging user experiences around the role of a
protagonist. For instance, much-hailed IS system Fac¸ade
situates users as a guest in the midst of a conﬂict between two
autonomous characters. The few video games that include IS
elements (e.g., ‘‘Fahrenheit’’ and ‘‘Heavy Rain’’) also have
user-control central game characters. IS prototype EmoEmma
introduces a different approach, labeled ‘ghost mode’. Here,
users do not play a story character, but observe and inﬂuence
ongoing story events by, for example, moving or adding
objects or giving commands to virtual actors. Like ghosts,
users move freely through the story environment and interact
with its elements. Ghost mode therefore comes close to being
a global inﬂuencer: a story cocreator who takes charge of
story developments without ties to an individual protagonist.
Theoretically, such different ways to involve users in IS are
likely to elicit different user experiences. Video game identi-
suggests that IS based on character
control should drive users’ sense of entertainment through
simulated self-experiences of being a character, living in a
story world, and feeling characters’ emotions. These simu-
lated self-experiences, in turn, likely elicit entertainment ex-
periences such as character identiﬁcation, presence, and
affect, that is, assuming identity, location, and inner states of a
In contrast, IS based on story control may elicit
user experiences dominated by a sense of control or auto-
as well as effectance and ﬂow,
since this global role,
along with its means of interaction, helps IS players’ pursue
their goal of creating a worthwhile story, and thus facilitates
perceptions of self-sufﬁciency and task-oriented attention.
The only empirical study conducted thus far on user re-
sponses to different user modes in IS compared usability-
oriented outcomes for actor and ghost mode Emo Emma
It found that users covered far larger distances
within the virtual story environment in ghost mode, sup-
porting the assumption of autonomy as a key dimension of
the user experience in this design approach.
The present study goes beyond measuring usability and
player behaviors—it instead compares entertainment experi-
ences of users of Emo Emma in both local and global modes,
employing a pre-established measurement battery
a broad range of measures derived from entertainment theory.
Based on the arguments above, we expect actor mode to
better facilitate experiences of presence, character identiﬁ-
cation, and affect, whereas ghost mode will better facilitate
experiences of autonomy, effectance, and ﬂow. Overall en-
joyment, as well as theoretically established drivers of media
enjoyment—curiosity, suspense, aesthetic pleasantness, sys-
tem usability, user satisfaction, character believability, and
pride—are tested in an exploratory fashion. In addition, we
will test gender differences in entertainment experiences. Prior
research suggests that for male players, the sense of control
facilitated by ghost mode might be more appealing,
whereas female players might be more attracted to the sense of
communicative involvement facilitated by the actor mode.
An experiment compared players’ responses to an inter-
active story played in an actor versus ghost mode. A total of
34 university students (11 men, 23 women; average age
M=22.0 years, SD =1.92 years) with a low-to-moderate de-
gree of computer game literacy (M=1.71, SD =0.84 on a scale
from 1 to 3) participated in the study. Comparisons between
playing modes were implemented within subjects. The order
in which participants interacted in the actor-versus-ghost
mode was balanced, counteracting possible order effects.
The IS stimulus system was Emo Emma, an advanced
prototype developed at the Teesside University, UK.
on the classic French novel Madam Bovary by Gustave
Flaubert, this system allows users to engage in a romantic
conversation between two characters, situated in a mansion.
In the actor mode, users play the role of Rodolphe Boulanger
(Emma Bovary’s admirer), who intends to express his ro-
mantic feelings toward Emma, in spite of her marriage. In the
ghost mode, players are bodiless and invisible in the virtual
environment, can observe the ongoing conversation among
Rodolphe and Emma, explore the house freely, manipulate
objects, and inﬂuence the behavior of either character. In both
usage modes, the scene lasts between 4 and 6 minutes.
Mouse, keyboard, and vocal commands serve as user inputs.
Upon arrival in the laboratory, participants received a
short training in interacting with Emo Emma for about
5 minutes. Next, half of the participants were ﬁrst exposed to
an IS sequence in the actor mode, whereas the other half was
ﬁrst exposed to the ghost mode. Subsequently, participants
completed a questionnaire consisting of 14 previously de-
veloped and validated
measures that capture a broad range
of drivers of media enjoyment: curiosity, suspense, ﬂow,
aesthetic pleasantness, enjoyment, affect, role adoption, sys-
tem usability, user satisfaction, character believability, effec-
tance, presence, autonomy, and pride. Experience dimensions
were measured on a 5-point Likert scale using two to ﬁve
items each. Internal consistency scores for all scales (Table 1)
were acceptable, except for negative affect whose two items
showed only a weak-to-moderate correlation, and suspense.
Subsequently, participants proceeded to interact with Emo
Emma in the opposite (ghost or actor) mode. Then, they
completed the questionnaire again. Finally, participants
received 20 EUR as compensation, were debriefed, and
Within-subject comparison of self-reported experiences
between playing in the actor-versus-ghost mode using paired
sample t-tests reveals that user experiences indeed differ be-
tween playing modes (see Table 1). When playing in the ghost
mode, participants reported signiﬁcantly higher levels of ef-
fectance, and marginally higher levels of autonomy and ﬂow.
These results conﬁrm that the ghost mode facilitates greater
degrees of empowerment to pursue storytelling goals. Pride
was also signiﬁcantly higher in the ghost mode. Providing
players with better means to pursue their goals seems to
improve the likelihood they experience a sense of accom-
plishment. More unexpectedly, curiosity was also marginally
higher in the ghost mode. Apparently, by offering a (nearly)
constraint-free mode of interaction, the ghost mode triggered
more user interest in the consequences of actions. Finally, user
PLAYING IN OR OUT OF CHARACTER 631
satisfaction was marginally higher in the ghost mode, indi-
cating that it better met players’ prior expectations about IS
Playing in the actor mode resulted in relatively low expe-
rience ratings on the mentioned dimensions, which suggests
that playing a predeﬁned role within the narrative yields
fewer perceptions of control and successful implementation
of intentions in users. Converse to our expectations, the actor
mode did not yield higher scores on presence, character
identiﬁcation, and affect. This contradicts prior research on
video game identiﬁcation
and indicates that perhaps the
immersive qualities of the current system did not (yet) elicit
perceptions of melting with a character.
An analysis of gender differences in user experiences using
a repeated measures ANOVA showed that while male users
overall enjoyed the ghost mode more than the actor mode
(M=3.94, SD =0.91 vs. M=3.74, SD =0.90), for female users it
was the other way around (M=3.18, SD =1.29 vs. M=3.59,
SD =0.92; F(1, 32) =4.57, p=0.04). This coincides with the
notion that men prefer the sense of control provided by the
ghost mode, while women prefer the communicative in-
volvement provided by the actor mode. This notion is not
backed up by more speciﬁc user experience measurements;
none revealed gender differences.
The current study tested whether different player modes
(actor mode, exerting local inﬂuence versus ghost mode, and
exerting global inﬂuence) yield different types of user expe-
riences (senses of immersion vs. control) in IS environments.
As expected, control-related experiences such as autonomy,
effectance, and ﬂow were higher in the ghost mode. Also, the
ghost mode triggered a higher sense of accomplishment, user
satisfaction, and curiosity. Contrary to our expectations,
playing in the actor mode did not improve involvement-
related experiences such as presence, character identiﬁcation,
We conclude that participants responded positively to the
greater level of goal-oriented control enabled by the ghost
mode. In contrast, playing in the actor mode seems to come
with perceived constraints with respect to storytelling goals.
The ghost mode offers more freedom and a broader arsenal of
possible user interventions that can be employed to exert a
more global inﬂuence. In addition, however, the ghost mode
seems to offer a detachment from story characters that may
serve to take away psychological constraints to pursue the
storytelling goal. As one participant put it, ‘‘It was easier to
play the ghost, because giving Rodolphe tips about what
to say to her was easier for me than actually say these things
to Emma in a convincing way.’’ The combined beneﬁts of
control, freedom, and character detachment may have
contributed to males’ greater enjoyment of the ghost mode;
female participants in contrast preferred the more socially
involving experience provided by the actor mode.
The present research implies that IS environments may
face a speciﬁc challenge with respect to user experiences, that
is, allowing users to exert global control over story develop-
ments while keeping them immersed in story developments.
One way to face this challenge is by producing extremely
convincing social interactions with digital agents, as well as
presence-evoking environments, to maintain suspension of
disbelief despite full story control. Surely, creating such en-
vironments poses an AI challenge for many forms of digital
entertainment, but for IS, where linear storylines and con-
straints to user autonomy are thrown overboard, it might be
Given the infant state of IS, in the near future, most gains
can be expected by focusing on an improvement of the
experience of global control,suchasinEmoEmma’sghost
mode. Such an improvement would make IS environments
stand out from other digital entertainment environments
(e.g., video games). Experiences such as freedom,
and successful implementation of one’s own
are established drivers of interactive enjoy-
ment, and could provide future IS environments with a
From a cyberpsychology perspective, our research implies
that technological drivers can have a relevant impact on IS
entertainment experiences. Future conceptual work could
focus on how technological and content-related factors (e.g.,
entertainment vs. education-oriented IS environments) in-
teract in shaping users’ experiences.
This research was funded by the European Commission
(Network of Excellence IRIS—Integrating Research on
Interactive Storytelling—FP7-ICT-231824). We thankfully
acknowledge the Commission’s support.
Author Disclosure Statement
No competing ﬁnancial interests exist.
Table 1. Comparison of User Ratings
Between Actor Mode and Ghost Mode
Actor mode Ghost mode
assessment) M SD M SD p
System usability a=0.61/0.70 4.11 0.80 4.22 0.65 0.28
User satisfaction n/a 3.09 1.06 3.35 0.98 0.08
Presence a=0.79/0.76 3.26 0.85 3.10 1.01 0.22
n/a 3.12 0.98 3.06 1.04 0.80
Effectance r =0.82/0.88 2.24 0.95 2.88 1.27 0.007
Autonomy a=0.84/0.87 2.17 0.85 2.47 1.08 0.05
Curiosity a=0.72/0.82 3.59 0.80 3.86 0.56 0.05
Suspense a=0.56/0.56 3.61 0.61 3.49 0.80 0.37
Flow a=0.62/0.71 3.09 0.71 3.31 0.79 0.09
a=0.81/0.87 2.33 0.91 2.44 0.94 0.34
Pride a=0.87/0.88 2.21 0.84 3.01 1.10 0.001
Enjoyment r =0.86/0.94 3.68 1.10 3.69 0.95 0.91
Affect: positive a=0.86/0.84 3.00 0.95 3.17 0.91 0.14
Affect: negative r =0.36/0.37 1.97 0.82 1.88 0.65 0.33
n/a 2.76 1.14 2.63 1.02 0.34
Marginal difference at p<0.1.
Signiﬁcant difference at p<0.05.
Reliabilities of scales with only two items were assessed using
Pearson’s r correlations. No reliability (n/a) is stated for one-item
632 ROTH ET AL.
1. Cavazza M, Lugrin JL, Pizzi D, et al. Madame bovary on the
holodeck: immersive interactive storytelling. ACM Multi-
media 2007; 651–660.
2. Cavazza M, Donikian S, Christie M, et al. (2008) The IRIS
network of excellence: integrating research in interac-
tive storytelling. In Spierling U, Szilas, N, eds. Interactive
storytelling 2008. Lecture notes in computer science 5334.
Heidelberg: Springer Berlin, pp. 14–19.
3. Murray JH. (1997) Hamlet on the holodeck: the future of narra-
tive in cyberspace. Cambridge: The MIT Press.
4. Tanenbaum J, Tanenbaum K. (2008) Improvisation and
performances as models for interacting with stories. In
Spierling U, Szilas N, eds. Interactive storytelling 2008. Lecture
notes in computer science 5334. Heidelberg: Springer Berlin,
5. Roth C, Vermeulen I, Vorderer P, et al. Exploring replay
value: shifts and continuities in user experiences between
ﬁrst and second exposure to an interactive story. CyberP-
sychology, Behavior, and Social Networking 2012; 15:
6. Klimmt C, Hefner D, Vorderer P. The video game experience
as true identiﬁcation: a theory of enjoyable alterations
of players’self-perception. Communication Theory 2009; 19:
7. Vermeulen I, Roth C, Vorderer P, et al. (2010) Measuring
user responses to interactive stories: towards a standardized
assessment tool. In Aylett R, Petta P, Riedl M, eds. Interactive
storytelling–proceedings of the 3rd international conference on
interactive digital storytelling 2010. Lecture notes in computer
science 6432. Amsterdam: Springer, pp. 38–43.
8. Vorderer P, Klimmt C, Ritterfeld, U. Enjoyment: at the heart
of media entertainment. Communication Theory 2004;14:
9. Mateas M, Stern A. Fac¸ade: an experiment in building
a fully-realized interactive drama. Game Developers
Conference, Game Design track 2003; 2:82.
10. Lugrin JL, Cavazza M, Pizzi D, et al. (2010) Exploring the
usability of immersive interactive storytelling. In the 7th
ACM symposium on Virtual Reality Software and Tech-
nology, 2010, pp. 103–110.
11. Bessiere K, Seay AF, Kiesler S. The ideal Elf: identity
exploration in world of warcraft. CyberPsychology and
Behavior 2007; 10:530–553.
12. Klimmt C, Hefner D, Vorderer P, et al. Identiﬁcation with
video game characters as automatic shift of self-perceptions.
Media Psychology 2010; 13:323–338.
13. Trepte S, Reinecke L. The pleasures of success: game-related
efﬁcacy experiences as a mediator between player perfor-
mance and game enjoyment. CyberPsychology, Behavior,
and Social Networking 2011; 14:555–557.
14. Ryan RM, Rigby CS, Przybylski A. The motivational pull of
video games: a self-determination theory approach. Moti-
vation and Emotion 2006; 30:347–365.
15. Brunner C, Bennett D, Honey M. Girl games and techno-
logical desire. In Cassell J, Jenkins H, eds. From Barbie to
Mortal Kombat: gender and computer games. Cambridge, MA:
MIT Press, pp. 72–88.
16. Jansz J, Tanis M. Appeal of playing online ﬁrst person
shooter games. CyberPsychology and Behavior 2007; 10:
17. Hartmann T, Klimmt C. Gender and computer games:
exploring females’ dislikes. Journal of Computer-Mediated
Communication 2006; 11:2.
18. Klimmt C, Hartmann T, Frey A. Effectance and control as
determinants of video game enjoyment. CyberPsychology
and Behavior 2007; 10:845–847.
Address correspondence to:
Center for Advanced Media Research Amsterdam (CAMeRA)
VU University Amsterdam
De Boelelaan 1105
Amsterdam 1081 HV
PLAYING IN OR OUT OF CHARACTER 633