University of Southern California Dissertations and Theses

Self-presence: body, emotion, and identity extension into the virtual self

(USC Thesis Other)

Self-presence: body, emotion, and identity extension into the virtual self

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

SELF-PRESENCE: BODY, EMOTION, AND IDENTITY EXTENSION INTO THE
VIRTUAL SELF

by

Rabindra Ayyan Ratan

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(COMMUNICATION)

August 2011

Copyright 2011 Rabindra Ayyan Ratan

ii
Epigraph

Self-Presence, a Limerick

One thing we don’t quite understand so far,
is the way people connect to their avatars.
So to get at this essence,
I propose we use self-presence,
a concept I hope becomes standard, yarrr!

- R. Ratan, recited at numerous presentations

iii
Dedication

This dissertation is dedicated to Jeanne and Atticus, my wife and son, for
giving me the strength, motivation, and opportunity to do my best on the serious stuff
and have fun the rest of the time.

iv
Acknowledgements

I would like to thank…
Dmitri Williams for holding me to high standards as my committee chair, as well as
for using his guild leader social capital to help me drum up survey responses.
Kwan Min Lee for encouraging my use of psychophysiological methods, as well as
for his generous support on my committee.
Michael Dawson, as my committee member, for sharing his psychophysiological
expertise and learning about a new, virtual area of application.
Jeremy Bailenson for planting the seed of this dissertation topic and sharing some
water along the way.
Béatrice Hasler for including the SPQ in her large-scale international studies and
sharing the data.
Christelle Williams for her enormous assistance running the experiment and wading
through and cleaning tons of data.
Christopher Courtney for sharing his psychophysiological expertise and advisor.
Adam Kahn, Brenda Osuna, and Scott Sanders for their superb statistical support.
All (four) of my parents, for never letting me give up on this academic thing.
And last but certainly not least, Olga Chandra, my mom, for countless hours of loving
editorial feedback not only on this dissertation, but since my first written sentence.

v
Table of Contents

Epigraph ii
Dedication iii
Acknowledgements iv
List of Tables vi
List of Figures viii
Abstract ix
Introduction: An Overview 1
Chapter 1: Building a Framework of Self-Presence 5
Chapter 2: Self-Presence in Collaborative Virtual Environments 23
Chapter 3: Self-Presence and Online Gamers 46
Chapter 4: Controlled Manipulations of Self-Presence via Avatar Assignment 58
Chapter 5: Physiological Correlates of Self-Presence During Avatar Use 79
Chapter 6: Physiological Effects of Self-Presence after Avatar Use 104
Chapter 7: Self-Presence and Math Performance 124
Conclusion: Summary and Future 137
References 146
Appendices:
Appendix A: Chapter 2 Appendix 157

Appendix B: Chapter 3 Appendix 160

Appendix C: Chapter 4 Appendix 162

Appendix D: Chapter 7 Appendix 163

vi
List of Tables

Table 1.1. Summary of the Self-Presence Framework: 16

Table 1.2. Self-Presence Questionnaire (SPQ): Update at time
of dissertation completion 19

Table 2.1. Pattern Matrix of SPQ 29
Table 2.2. Structure Matrix of SPQ 29
Table 2.3. Pearson correlations 29
Table 2.4. Pattern Matrix of SPQ 33

Table 2.5. Structure Matrix of SPQ 33

Table 2.6. Pattern Matrix of Extended SP 33
Table 2.7. Structure Matrix Extended SP 33
Table 2.8. Pearson correlation between SPQ factors and other variables. 35
Table 2.9. Extended SP Pattern Matrix 37

Table 2.10. Extended SP Structure Matrix 37

Table 2.11. Final SPQ Pattern Matrix 37

Table 2.12. Final SPQ Structure Matrix 37

Table 2.13. Pearson correlation between SPQ factors, and other variables. 38

Table 2.14. SPQ descriptives by avatar type 40
Table 3.1. Pattern Matrix of SPQ 53

Table 3.2. Structure Matrix of SPQ 53

Table 3.3. Pearson correlations between self-presence levels 54
Table 3.4. Descriptive statistics for MMO players by sex 54

Table 3.5. Descriptive statistics for male players separated by game genre 54

Table 3.6. Cronbach's alphas for self-presence measures segmented by
participant sex and game genre 54

vii

Table 4.1. Pattern Matrix of SPQ 67

Table 4.2. Structure Matrix of SPQ 67

Table 4.3. Pearson correlation between composite measures of self-presence 68
Table 4.4. Regression Models on Extended (identity) Self-Presence 70
Table 4.5. Regression Models on Proto (body) Self-Presence 71
Table 4.6. Regression Models on Core (emotion) Self-Presence 71
Table 4.7. Estimated Marginal Means of Proto Self-Presence by Condition 71

Table 5.1. Regression Models on Heart Rate 95

Table 5.2. Regression Model on Electrodermal Activity 96

Table 5.3. Regression Model on Zygomatic EMG 96

Table 5.4. Regression Model on Corrugator EMG 97

Table 5.5. EDA descriptives by condition (range corrected scores) 98
Table 5.6. Corrugator EMG descriptives by condition (millivolts * 100) 98

Table 6.1. Regression Models on Tonic Arousal Measures 112

Table 6.2. Regression Models on Tonic Emotional Valence Measures 113

Table 6.3. Regression Models on the Phasic HR Metric for Hits 1-3 114
Table 6.4. Regression Models on the Phasic EDA Metric for Hits 1-3 115
Table 6.5. Regression Models on Phasic Zygomatic EMG H1-3 116
Table 6.6. Regression Models on Phasic Corrugator EMG H1-3 117
Table 6.7. Means of Phasic EDA Change after Hit 1 by
Avatar Assignment Condition 122

Table 7.1. Regression Models on Number of Correct Math Responses 132

viii
List of Figures

Figure 1.1. The Three Levels of Self 9
Figure 2.1. Seminar feedback session in a mixed-reality
setting with video-conferencing and a 3D virtual world. 26

Figure 2.2. International virtual team meeting in UNIworld, the
3D virtual environment used for student collaboration. 27

Figure 4.1. Avatar assignment examples for the experimental conditions. 63

Figure 4.2. Estimated Marginal Means of Proto Self-Presence by Condition 72

Figure 5.1. Electrode placement on the arm and face 88
Figure 5.2. Example of raw physiological output 89
Figure 5.3. Example of transformed physiological output 89
Figure 5.4. Corrugator EMG scatterplot 99

ix
Abstract

Our media landscape increasingly facilitates the creation and use of virtual
self-representations, i.e., avatars. As a result, there is a growing body of research that
focuses on avatar use and its effects. The present dissertation aims to contribute to
and standardize this research by developing a conceptual framework that describes the
ways people connect to their avatars on three neurologically-defined levels of self:
body, emotion, and identity. The concept of self-presence is at the heart of this
framework and is this dissertation’s main focus. After carefully defining self-
presence and offering its operationalization, Self-Presence Questionnaire (SPQ), the
majority of the dissertation is dedicated to examining self-presence and the SPQ
within a variety of virtual contexts, from collaborative student-oriented virtual
environments to multiplayer online games to a movement-based console game.
Results from five studies suggest that self-presence is associated with numerous
constructs, including game genre, physiological responses to avatar use, math task
performance after avatar use, and avatar customization, personalization, and gender.
Overall, the dissertation supports the claim that self-presence is an internally
consistent, reliable, valid, and valuable concept for avatar-use research, but there are
still many open questions regarding the nature of self-presence. Research in this area
may eventually lead to a full theory of self-presence, thereby contributing to a better
understanding of our evolving media landscape and perhaps even our selves.

1
Introduction: An Overview

Our media landscape increasingly facilitates the creation and use of virtual
self-representations, i.e., avatars. As a result, there is a growing body of research that
focuses on avatar use and its effects. The present dissertation aims to contribute to
this research by providing a theoretical framework and operational tool to aid in the
examination of avatar use. This framework and tool are described in the first chapter,
tested for internal consistency and validity tested in the following five chapters,
applied within an avatar use research context in the penultimate chapter, and directed
toward potential future applications in the conclusion. This introduction summarizes
each chapter in more detail and emphasizes the dissertation’s main contributions.
Chapter 1 describes how our developing media landscape increasingly
facilitates new types of virtual self-representations (i.e., avatars), which has spawned
a growing body of research dedicated to avatar use. This research relies on the
assumption that people feel connected to their avatars in some way, but there is no
standardized concept for describing such connections. In order to fill this need, the
present dissertation offers the framework of self-presence, which describes the way
people connect to their avatars on three levels: body, emotion, identity. After
thoroughly explicating this framework, the chapter concludes by describing the Self-
Presence Questionnaire (SPQ), a measurement tool that can be used to operationalize
each level of self-presence. The remainder of the dissertation is dedicated to
examining self-presence and the SPQ with the intent of establishing their importance
within avatar use research.
Chapters 2 and 3 describe tests of the SPQ within numerous virtual contexts
and populations. The results from these tests provide support for the internal

2
consistency and validity of the SPQ as well as the concept of self-presence in general.
One notable finding is that avatar customization and personalization (using a face
picture) relates to the different levels of self-presence in different ways. This suggests
that the distinction between customization and personalization deserves attention.
Overall, the results from these chapters suggest that self-presence is a valuable
construct across various virtual contexts with diverse types of virtual self-
representations.
Chapters 4 through 7 examine self-presence within an experimental context,
providing more control and detailed measurement than the previous chapters. These
chapters focus on different facets of the same experiment. This experiment included a
manipulation of avatar customization as well as gender consistency. Half of the
participants used an avatar they customized, and half of the participants used a
consistently gendered avatar. The virtual environment was a Nintendo Wii cartoonish
sword-fighting game. Physiological data (arousal and emotional valence) was
recorded from the participants during and after they played the game.
Chapter 4 focuses on the relationship between the avatar assignment
conditions and responses to the SPQ, which the participants provided immediately
after playing. Results mostly support expectations for relationships between the
individual levels of self-presence and avatar customization/gender consistency. Some
of the unexpected findings suggest that future research should avoid certain nuances
in the experiment design and also should pursue the customization/personalization
distinction, as noted in Chapter 2.
Chapter 5 focuses on the physiological correlates of self-presence while
participants were playing the game. Some of the results are consistent with
expectations regarding self-presence, but overall the results are clearer for the

3
measures of emotional valence than arousal. This makes sense because the latter is
more likely to be obscured by physical movement. Regardless of this issue, the
results are purely cross-sectional because self-presence and the physiological
responses both occur during game play.
Chapter 6 attempts to address this issue of causality by focusing on the
participants’ physiological responses while watching their avatar get attacked after
initial avatar use and responses to the SPQ. By controlling for previous physiological
responses, these tests are able to determine the causal effect of self-presence on
subsequent physiology. Results are consistent with expectations regarding the
different rates at which the different levels of self-presence dissipate after
disconnection from an avatar. Overall, these results provide the strongest support for
the framework’s validity within the dissertation.
Chapter 7 is unique from the previous chapters in that it does not focus on self-
presence itself but instead attempts to apply the concept of self-presence to a larger
topic, learning in video games. To this end, the experiment attempted to induce
avatar-gender-related stereotype threat before participants performed a mental math
task, i.e., participants were told that scores would be compared between people who
used male and female avatars. Unfortunately, this treatment was not strong enough to
induce stereotype threat and so the chapter does not provide support for the
expectation that self-presence plays a role avatar-related stereotype threat. However,
one notable finding is that one of the levels of self-presence was negatively related to
math performance. The chapter describes both state and trait-based interpretations of
this finding as well future research that could further explore it.
In Chapter 8, the dissertation concludes by summarizing the contributions and
questions raised by this dissertation and then suggesting various approaches to

4
research that can benefit from and contribute to an understanding of the concept of
self-presence. These approaches may focus on behavioral effects of avatar use, such
as research on the Proteus Effect. Or these approaches may focus on psychological
effects, such as research on representations of possible selves via avatars.
Applications of self-presence may also help with the reinterpretation of previous
findings or the standardization and thus comparison across avatar use research. Self-
presence may also contribute to research on phenomena that are not commonly
associated with avatars but to which avatar use may contribute, such as learning-
oriented video games. The chapter then suggests that there are many other types of
research that may benefit from the application of self-presence and that regardless of
focus, such research is likely to also contribute to a greater understanding of self-
presence. Concluding the dissertation, the chapter reflects on the general value of a
full theory of self-presence in our evolving media landscape.

5
Chapter 1: Building a Framework of Self-Presence

Chapter 1: Abstract
The framework of self-presence describes how people connect to their virtual
self-representations on three distinct levels of self (body, emotion, identity). The
present chapter draws from the fields of presence and neuroscience to describe the
foundations of the concept of self-presence, the three-level framework of self-
presence, and a measurement tool for this concept, the Self-Presence Questionnaire.
It should be noted that much of the content in this chapter comes directly from papers
previously written by the author (Ratan, 2010, 2011).

Self-representations are an important facet of human existence. We pick
names for our children that we hope will represent them well, craft our CVs to
highlight significant accomplishments, and choose images for profile pictures so that
others may recognize us. Most people are connected to numerous self-
representations, but the significance of these connections differs between individuals.
For example, some people may feel a close affinity to their given birth names while
others change their names as adults. Specific types of self-representation may also
affect the ways a person connects to them. For example, some people may feel that
their CVs only highlight career-related aspects of identity while their Facebook
profiles present a truer depiction of the self.
Our present media landscape offers a huge variety of potential self-
representations, from 3-D anthropomorphic avatars that mimic users’ body
movements to simple usernames in text-based chat rooms. This dissertation focuses
more on the former than the latter, offering a framework that describes the ways

6
people connect to virtual self-representations, and thus serves as a potential tool for
the growing body of research on the effects of using avatars. Research in this area
includes studies that examine the effects of various avatar-related attributes, such as
avatar development guidelines (Jin, 2009) and avatar characteristic-based behavioral
expectations (Yee, Bailenson, & Ducheneaut, 2009), on numerous outcomes, such as
healthy eating (Fox, Bailenson, & Binney, 2009), exercise (Fox & Bailenson, 2009),
racial bias (Groom, Bailenson, & Nass, 2009), aggression and group cohesion (Pena,
Hancock, & Merola, 2009), and body size judgments (Chandler, Konrath, & Schwarz,
2009). All of this research assumes that people feel some sort of connection to their
avatars, but there is no standardized concept for understanding such connections. The
framework offered in this dissertation can contribute to the strength and
understanding of these effects by allowing researchers to control for differences in the
extent to which people feel connected to their avatars.
The concept at the heart of this framework is called self-presence, which can
be broadly described as the extent to which the self is present (relevant) during media
use. This chapter draws from the fields of presence and neuroscience in order to
explicate the concept of self-presence and then offers a measurement tool for the
concept.
Presence as Precursor
The concept of self-presence is closely related in origin and meaning to the
concept of presence. Thus, the present explication of self-presence begins in the field
of presence. Because this field extends into many disciplines and discussions, only
those works that provide direct foundations for the current formulation of self-
presence are examined here.

7
Presence, as a concept within the field of Communication, grew out of the
need to understand the effects of technologies and media that increasingly embody the
user, namely robotics and virtual reality. Telepresence, a term adopted by Steuer
(1992) from various previous uses to mean the experience of being there in a
mediated environment, is introduced as an integral facet of virtual reality technology.
He goes on to describe the technological dimensions that contribute to telepresence,
including the vividness and interactivity of a medium. Biocca (1997) adopts the
shorter term presence to mean the perception of being there in a physical, mediated,
or imagined space. He describes how virtual reality technologies aim to induce the
feeling of physical presence through sensory and motor engagement. Lombard &
Ditton (1997) define the term presence as the “perceptual illusion of nonmediation”,
thus restricting the use of the term only to media environments. They describe the
causes of presence, most of which are characteristics of the medium or other social
actors in the medium. Lee (2004) defines presence as a psychological state in which
virtual objects, social actors, or virtual selves are experienced in sensory and non-
sensory ways. He clarifies that virtual objects, social actors, or virtual selves can be
either para-authentic, having real-world correlates, or artificial, without real-world
correlates. While these four definitions differ significantly, they all imply that
presence occurs when mediated experiences are treated as nonmediated experiences
in some way.
Another important perspective on presence that is relevant to the current
formulation of self-presence comes from the field of neuroscience. This work (Riva,
Waterworth, & Waterworth, 2004; Waterworth & Waterworth, 2003) adopts
Damasio’s framework of consciousness and self (1994, 1999), which opposes
Descartes’ 350 year-old dualist notion of the mind as a separate entity from the

8
machine-like body. Damasio (1994) argues that such dualism is incorrect and instead
that the mind and body are always integrated, by biochemical and neural circuits,
when humans interact with their environment. He offers neurological evidence for
this argument and further stresses that the experience of emotions and cognitions are
embedded in the body’s biological processes. Damasio (1999) provides a framework
of the self that incorporates this interrelation between the mind and body. This
framework is comprised of the two levels of consciousness that govern human
experience. The first level, core consciousness, arises through interactions with
objects, i.e., neural patterns that represent facets of the immediate environment. The
second level, extended consciousness, arises when the lived past and anticipated
future are recorded in memory and then reactivated in order to generate a sense of self
knowing. While core consciousness governs action in the moment, extended
consciousness governs the management of all information that is relevant to future
action, including identity and individual perspective.
Waterworth and Waterworth (2003) use this framework to argue that presence
is an evolved mechanism that allows humans to distinguish between internal and
external worlds. The former are facilitated by extended consciousness and the latter
are facilitated by core consciousness and are usually more relevant to immediate
survival. Based on this reasoning, they claim that media, such as virtual reality
technologies, are able to induce a state of presence by “tricking” core consciousness
into engaging with virtual objects as if they were physical objects in the immediate
environment.
Damasio (1999) elucidates the relationship between core and extended
consciousness by defining three distinct levels of self. First, the proto self is an
organism’s sense of physical being, based on neural patterns that map the physical

9
structures of the organism, of which the organism is not conscious. Next, the core self
is an organism’s emotional state (at the fundamental physiological level), of which the
organism may be conscious, generated through encounters between the proto self and
objects in the environment. And last, the autobiographical self is an organism’s
conscious idea of self, or identity, that is dependent on personal traits and memories
of past experiences. Linking the levels of self with the levels of consciousness,
Damasio (1999) writes, “The nonconscious neural signaling of an individual organism
begets the proto-self which permits core self and core consciousness, which allow for
an autobiographical self, which permits extended consciousness” (p. 230). In other
words, the embodied biological self is intrinsically linked to personal and social
identity of the self. An illustration of these three levels of self is depicted in Figure
1.1.
Figure 1.1. The Three Levels of Self

Riva, Waterworth, & Waterworth (2004) define three levels of presence that
are associated with Damasio’s (1999) three levels of self. Proto presence is an
organism’s ability to couple perceptions and movements and thus differentiate itself

10
from the external world. Core presence is an organism’s ability to focus sensorial
experience on present tasks and relevant objects in the environment. Extended
1

presence is an organism’s ability to recognize the significance of previous
experiences as they relate to the self. Over the course of human evolution, each of
these levels of presence would have been important to survival in different ways.
Proto presence is necessary for an organism to utilize its body and thus navigate the
dangers and opportunities of an environment. Core presence is necessary to
determine the relative importance of objects in the environment to survivability.
Extended presence is necessary to effectively learn from the past and plan for the
future.
Riva, Waterworth, and Waterworth (2004) argue that media can potentially
induce each level of presence, though the extent of media complexity required in
order to do so is highest for proto presence and lowest for extended presence. In
order for a medium to induce proto presence, the medium must include advanced
proprioceptive inputs and outputs that mimic natural interactivity and provide rapid
response times. There is no current virtual reality technology that can achieve such
integration with the body. Core presence can be induced by less complex media that
present objects that seem realistic and relevant to the core self. Even media as simple
as television can induce this level of presence. Inducing extended presence requires
only intellectually or emotionally significant content to be transmitted via a simple
medium (e.g., text).
This application of Damasio’s (1999) framework provides an important
perspective on the concept of presence, which relates to the experience of mediated

1
They use extended instead of autobiographical to refer to extended consciousness, not just
autobiographical self. The present dissertation adopts this terminology.

11
environments in general. But given that this framework focuses on the concept of the
self, it also has the potential to contribute distinctively to our understanding of
mediated self-representations. The present chapter fulfills this opportunity by
following Riva, Waterworth, and Waterworth’s (2004) example, but instead of
applying Damasio’s (1999) framework to presence, applies it to self-presence.
Putting the Self in Presence
As mentioned earlier, the concepts of self-presence and presence share origins.
Essentially, though in different terms, Biocca (1997) and Lee (2004) describe both
concepts as psychological states in which the virtual is treated as real, to some extent.
Presence relates to virtual environments while self-presence relates to virtual self-
representations. Just as the experience of presence involves engaging with a virtual
environment as if it were a physical environment, the experience of self-presence
involves using a virtual self-representation as if it were integrated with the physical
self. Specifically, Biocca (1997) introduced the term self-presence to describe a
user’s mental model of her body, physiological states, emotional states, perceived
traits, and identity within a virtual world. Lee (2004) used the term to mean a
psychological state in which virtual selves are experienced as the actual self in
sensory and non-sensory ways, referring mostly to identity but also mentioning body
schema.
Although the concept of presence has received much attention and refinement,
the concept of self-presence has not and thus has suffered from the lack of a standard
definition. Empirical research has treated it in different ways. For example, Ratan,
Santa Cruz and Vorderer (2007) examine self-presence as it relates to identity, while
Jin and Park (2009) examine the concept as it relates to physical location and
movements. The present paper addresses this need for a standard definition by

12
applying Damasio’s (1999) framework to the definitions introduced by Biocca (1997)
and Lee (2004).
Damasio’s (1999) framework of self maps easily onto the definition of self-
presence. Proto self relates to body schema, core self relates to emotional states, and
extended self relates to perceived traits and identity. Thus, this dissertation offers the
definition of self-presence as “the extent to which some aspect of a person’s proto
(body-schema) self, core (emotion-driven) self, and/or extended (identity-relevant)
self is relevant during media use.” The three levels of self-presence introduced by
this definition are perhaps more important than the concept of self-presence as a
whole. Each level is described in more detail below and summarized in Table 1.1.
Proto self-presence, which can also be called body-level self-presence, is
defined as “the extent to which a mediated self-representation is integrated into body
schema.” This definition builds on research about body schema integrating physical
objects, such as faux body parts or physical tools (Armel & Ramachandran, 2003;
Berti & Frassinetti, 2000; Botvinick & Cohen, 1998; Ehrsson, Spence, & Passingham,
2004; Farnè & Làdavas, 2000; Iriki, Tanaka, & Iwamura, 1996; Maravita & Iriki,
2004; Tsakiris & Haggard, 2005), as well as virtual or mediated body parts
(Ijsselsteijn, de Kort, & Haans, 2006; Iriki, Tanaka, Obayashi, & Iwamura, 2001;
Slater, Perez-Marcos, Ehrsson, & Sanchez-Vives, 2008). When such integration of a
virtual body part occurs, the individual interacts with virtual objects without
considering the mediation involved in the interaction. The experience of proto self-
presence involves using a virtual object as if it is an extension of the body. This
requires significant comfort or experience with the interface used to control the virtual
object. For example, people who experience high levels of proto self-presence control
their avatars with the same ease using a pen after years of practice. Just as the pen is

13
integrated into the hand’s movements with little thought, the avatar becomes part of
body schema, like an anthropomorphized fingertip.
Core self-presence, which can also be referred to as emotion-level self-
presence, is defined as “the extent to which mediated interactions between a self-
representation and mediated objects cause emotional responses.” The term emotional
here refers more to physiological reactions than affective states, though the two are
often related (Hugdahl, 1996). Everyday interactions between an individual’s body
and objects in the environment yield these responses (Damasio, 1999), such as
frowning when looking at something unpleasant. Armel, & Ramachandran (2003)
found that people responded emotionally (indicated by electrodermal activity) when a
faux body part associated with body schema was (falsely) injured. Similarly, if a
virtual self-representation is integrated into body schema (i.e., proto self-presence),
then interactions between this virtual self-representation and virtual objects should
also yield emotional responses. This experience is core self-presence.
Extended self-presence, which can also be called identity-level self-presence,
is defined as “the extent to which some aspect of a self-representation is related to
some aspect of personal identity.” There are numerous aspects of self that comprise
identity, and while some them are immutable, many can be intentionally emphasized
or curtailed (Goffman, 1959). This fluidity of identity means that people can choose
from numerous “possible selves” (Higgins, 1987; Markus & Nurius, 1987), especially
during computer-mediated communication (Bargh, McKenna, & Fitzsimmons, 2002;
Gibbs, Ellison, & Heino, 2006; Zhao, Grasmuck, & Martin, 2008). Mediated
environments allow for an extent of identity experimentation that is not easily
attainable offline (Turkle, 1995). For example, although most virtual world users
choose avatars that are consistent with their biological sex (Huh & Williams, 2010;

14
Hussain & Griffiths, 2008), many choose to switch genders (Roberts & Parks, 1999).
Research has found that women who choose male characters are more likely to
exhibit more masculine behaviors online, such as player-versus-player combat, than
feminine behaviors, such as chatting (Huh & Williams, 2010). This illustrates the
experience of extended self-presence with respect to gender. Those women for whom
masculinity is a salient identity characteristic extend this aspect of the self into their
self-representations.
The three levels of self-presence, proto, core, and extended, should be
considered separately, but just as Damasio (1994) argues that mind/body dualism is
false, they should also be considered as interrelated experiences. Further, just as in
Damasio’s (1999) framework of the self, the three levels should be hierarchically
related to some extent, with proto self-presence at the foundation, followed by core
self-presence, followed by extended self-presence. This does not mean that a lower
level is required to experience a higher level. In fact, the opposite may be true, just as
with Riva, Waterworth, and Waterworth’s (2004) framework, in which proto presence
is the most difficult to achieve. For example, extended self-presence may be
experienced more easily than proto self-presence because the former is achieved
through simple identity information input while the latter requires extreme comfort
with a media interface.
The temporal facets of each level of self-presence are also worth
consideration. Proto self-presence occurs only during avatar use. In the same way
that a false body part must be appropriately positioned in order for an individual to
integrate it into body schema (Armel & Ramachandran, 2003; Tsakiris & Haggard,
2005), some existing part of the body must be engaged in controlling the avatar in
order for the user to integrate this avatar into body schema. Thus, once the user

15
relinquishes control of the avatar, proto self-presence ceases within seconds if not
immediately. Similarly, core self-presence arises during the process of interaction
between the avatar and the virtual environment. However, the feeling of core self-
presence may continue after the user stops controlling the avatar. For example, if
someone who is experiencing a high degree of core self-presence while playing a
video game puts the controller down and takes a step back from the screen, any
positive or negative events that occur to the avatar may still elicit consistent emotional
responses from the individual. Thus, unlike proto self-presence, which diminishes
immediately after avatar use, core self-presence can have a residual effect that likely
dissipates over minutes or hours. Lastly, extended self-presence also arises during
avatar use, but can persist for much longer after avatar use than the other two levels of
self-presence. When someone customizes an avatar to reflect certain aspects of
identity, this customization will not change after avatar use. Thus, the extent to which
extended self-presence diminishes is related to the rate at which the avatar user’s
feelings of identity change. This rate tends to vary depending on an individual’s stage
of development, but at the very least is on a scale of days if not months (Erikson,
1959). These temporal differences between the three levels of self-presence will be
examined in Chapter 6 of this dissertation.

16
Table 1.1 Summary of the Self-Presence Framework

Proto Self-Presence Core Self-Presence Extended Self-
Presence
Also referred to
as
Body-level
self-presence
Emotion-level
self-presence
Identity-level
self-presence
Definition The extent to
which a mediated
self-representation
is integrated into
body schema
The extent to which
mediated interactions
between a self-
representation and
mediated objects
cause emotional
responses
The extent to
which some
aspect of a self-
representation is
related to some
aspect of personal
identity
High self-
presence
implies…
Self-
representation
treated as an
extension of the
body without
conscious
consideration of
the media
interface

Interactions between
self-representation
and mediated objects
cause strong
emotional responses
Self-
representation
reflects important
aspects of
personal identity
Temporal facets Arises and
persists only
during avatar use,
dissipates
immediately or
seconds
afterwards
Arises during media
use, may linger
afterwards for
minutes or hours
Arises during
avatar
customization and
use, dissipates at
rate of personal
identity change –
days, months,
years.
Sample
Questionnaire
Items
When using your
avatar, to what
extent do you feel
like you can reach
into the
game/virtual
environment
through your
avatar?
When scary events
happen to your
avatar, to what extent
do you feel afraid?
To what extent is
your avatar’s
gender related to
some aspect of
your identity?

The Importance of Self-Presence

The concept of self-presence may contribute to the development of a larger
understanding of the ways that our increasing use of mediated self-representations

17
affects our experience in the world. Damasio’s (1994) view of the self as a highly
integrated brain/body system implies that subjective experience is not just a function
of abstract thought but is largely dependent upon the structures within which such
thought takes place. This is consistent with Nagel’s (1974) assertion that in order for
a human to accurately assume the perspective of a bat, or any other non-human
organism, the human would have to take on the physical structures through which the
bat experiences life. Humans have modified their structures of interaction with the
world through technology and thus the subjective human experience has changed
along with the technologies of the time. For example, centuries ago, the clock served
as the dominant metaphor for the human condition, but was eventually replaced by the
steam engine, and more recently by the computer (Bolter, 1984). These technological
developments undoubtedly affected humans’ perceptions of time and distance. Some
cognitive theorists argue that we have entered a state of posthumanism in which the
human can be considered a distributed cognitive system comprised of the body and
brain’s intelligence, intelligent machines, and interfaces between them (Hayles,
1999). Virtual self-representations present a new layer of technology currently
transforming the human experience. The concept of self-presence helps to describe
the interface between people and their self-representations and thus contributes to our
understanding of this transformation.
In order for the concept of self-presence to achieve this goal, it must be
applied concretely within appropriate research contexts. Although self-presence
could ostensibly describe the ways that the self is relevant during the manipulation of
physical self-representations, such as hand puppets, this discussion of self-presence is
restricted to virtual self-representations because the present chapter recommends self-
presence as a tool for research on digital communication technologies. There are a

18
growing number of people communicating through such technologies, including
three-dimensional virtual worlds and massively multiplayer online games (MMOs),
and these media have the potential to embody users physically as well as emotionally,
psychologically and socially (Chan & Vorderer, 2006; Tamborini & Skalski, 2006).
Researchers in this area employ a variety of perspectives and techniques in order to
understand the effects of such technology use. For example, Yee, Bailenson, &
Ducheneaut (2009) found that participants in an immersive virtual environment who
used more attractive avatars were more confident in social interactions than
participants with less attractive avatars. Using a different type of virtual environment,
the MMO World of Warcraft, Lim and Reeves (2009) found that participants who
chose their own character experienced higher physiological arousal during game play
than those who were assigned a character. The experience of self-presence may
influence such effects and thus would be useful to this area of research.
Measuring Self-Presence
The present chapter offers an operationalization of self-presence, the Self-
Presence Questionnaire (SPQ), that relies on self-report and includes questions that
gauge the extent to which media use is relevant to an individual’s proto self, core self,
and extended self. It should be noted that the SPQ has been developed iteratively
throughout the stages of this dissertation project. During this process, it became
apparent that instead of serving as a concrete set of questions, the SPQ should provide
a template for questions that can be adapted appropriately to any given virtual
environment by a researcher. The most updated version of the SPQ, which reflects
this thinking, is listed in Table 1.2. Former iterations of the SPQ are provided in the
chapter appendices of the subsequent chapters.

19
Table 1.2. Self-Presence Questionnaire (SPQ): Update at time of dissertation
completion

Proto
Self-
Presence

1. When playing the game, how much do you feel like
your avatar is an extension of your body within the
game?
a. Response options: not at
all/somewhat/moderately/very
much/absolutely
2. When playing the game, how much do you feel your
avatar is a part of your body?
3. When using your avatar, to what extent do you feel
like you can reach into the game through your avatar?
4. When using your avatar, to what extent do you feel
like your arm is elongated (changed to “stretched” in
Study 2) into the game through your avatar?
5. When playing the game, to what extent do you feel
like your hand is inside of the game?

Core
Self-
Presence

1. When happy events happen to your avatar, to what
extent do you feel happy?
1. Response options: not at
all/somewhat/moderately/very/extremely
2. “happy events” can be substituted with
specific virtual interactions, such as “your
avatar killed a monster”
3. Relevant emotions can be selected according
to context: sad, surprised, disgusted, aroused,
scary/afraid, upset/angry

Extended
Self-
Presence

1. To what extent is your avatar’s appearance related to
some aspect of your personal identity?
1. Response options: not at
all/somewhat/moderately/very
much/absolutely
2. “appearance” can be substituted with other
relevant characteristics, including: name, race,
clothing, profile information, gender, skin
color, and hairstyle

Proto self-presence is measured by questions about the integration of the self-
representation into body schema. While this level of self-presence may occur beneath

20
conscious thought, which implies that self-report about related phenomena may not be
relevant, previous research has found that such subjective responses are valid.
Specifically, self-reported ratings of the integration of false body parts into body
schema were found to be positively related to neural activity in the premotor cortex,
where body schema is processed (Ehrsson, et al., 2004). Thus, self reported proto
self-presence is likely to reflect neural activity that occurs beneath conscious thought.
Drawing from research on this topic (e.g., Botvinick & Cohen, 1998), questions are
developed to gauge the extent to which a self-representation is considered to be an
extension of the individual’s body. Various phrasings are used to state this concept in
different ways. All of the questions imply a reduction of the separation between the
virtual and the physical body. This reduced separation may be manifested within the
game environment s or within the physical environment (e.g., you feel as though the
virtual self-representation is part of your physical body). An example of such a
question is, “When using your avatar, to what extent did you feel like you could reach
into the game through your avatar?”
For core self-presence, questions address how mediated interactions between
self-representations and virtual objects relate to emotional responses. These questions
are directed toward a handful of basic emotions (Ekman, 1992) that people could
potentially experience when using self-representation in a virtual environment or
video game, such as happiness and surprise. Because people respond emotionally to
media that do not include self-representations (Lee & Jung, 2005), the phrasings of
the questions clearly state that the virtual self must be active within the environment,
implying that the individual should consider the interaction between the virtual self
and virtual objects. This ensures that the questions are targeted at self-presence, not
presence in the general virtual environment. An example of a core self-presence

21
question is, “When scary events happen to your avatar, to what extent do you feel
afraid?” An adaptation of this question may refer to a specific virtual interaction and
another relevant emotion, such as “When your avatar was hit by an opponent, to what
extent did you feel surprised?”
In order to measure extended self-presence, questions address the extent to
which some aspect of a self-representation is related to some aspect of the
individual’s identity. Various aspects of identity that could potentially be represented
in a virtual environment or video game, such as gender and race, are included in these
questions. The formulation of these questions considers that identity often depends on
context (Goffman, 1959; Higgins, 1987; Markus & Nurius, 1987) and can be
expressed quite differently online and offline (Bargh, et al., 2002; Gibbs, et al., 2006;
Turkle, 1995; Zhao, et al., 2008). Therefore, the questions imply that there can be a
relationship between a self-representation and identity without the two being similar.
In other words, individuals may not express certain aspects of their identity in non-
mediated social situations, but they may express them through their avatars. The
phrasings of the extended self-presence questions attempt to account for such
potential discrepancies. An example of an extended self-presence question is, “To
what extent is your avatar’s appearance related to some aspect of your identity?” An
adaptation of this question may refer to another potentially relevant aspect of identity,
such as “To what extent is your avatar’s gender related to some aspect of your
identity?”
Testing the SPQ
The goal of this dissertation is to establish self-presence as a critical concept
within this growing area of research (e.g., Chandler, et al., 2009; Fox & Bailenson,
2009; Fox, et al., 2009; Groom, Bailenson, et al., 2009; Jin, 2010; Jin, 2009; Pena, et

22
al., 2009; Peng, Lee, & Heeter, 2010; Yee, et al., 2009). To this end, the remainder of
this dissertation is dedicated to developing the Self-Presence Questionnaire as a
standard measure of self-presence that is reliable and valid in a variety of virtual
contexts. Although this dissertation cannot include every possible test of the concept,
the results that will follow indicate that self-presence may serve as a common tool for
researchers in this area facilitating control and measurement of the ways and extent to
which people feel connected to their avatars.

23
Chapter 2: Self-Presence in Collaborative Virtual Environments

Chapter 2: Abstract
The present chapter examines data from a pilot study (N=31) and two larger
studies (N = 81 and N = 159) to test the internal consistency and validity of the Self-
Presence Questionnaire (SPQ). All three studies utilized similar methodologies, with
surveys administered to students who used virtual environments to collaborate on a
course project. The second study improved on the first by examining a larger sample
and the third improved on the second by also increasing sample size as well as adding
an avatar customization option that facilitated the comparison of customization and
personalization. Factor analyses across the three studies supported the SPQ’s
internally consistency. Validity tests were conducted by comparing individual SPQ
factors to relevant constructs, such as avatar customization. It should be noted that
much of the content in this chapter comes directly from existing papers (Ratan &
Hasler, 2010; Ratan & Hasler, 2009). The second author of these papers was
responsible for running the studies, but the conceptual framework, writing (aside from
the study details), and analysis come from the author of this dissertation.

The concept of self-presence cannot be accepted as a standard tool for research
on virtual self-representations until the internal consistency and validity of the
framework and operationalization have been established. To determine internal
consistency, factor analysis of the Self-Presence Questionnaire (SPQ) must result in
factors that are consistent with the three theorized levels of self-presence (i.e., proto,
core, and extended). Thus, as the framework of self-presence suggests, it is
hypothesized that the factors of self-presence are both distinct and interrelated.

24
Hypothesis 1: The levels of self-presence (proto, core, and extended) form
distinct but interrelated dimensions.
Previous research has found that males and females experience presence
differently, though the direction of this difference is not consistent. In some cases,
males report more presence (Lachlan & Krcmar, 2011), while in others, females
report more presence (Nah, DeWester, & Eschenbrenner, 2010). This discrepancy
may relate to the specific type of virtual environment or medium used, given that
females have been found to experience more presence than males when viewing large
screens but less than males when viewing small screens (Lombard, Reich, Grabe,
Bracken, & Ditton, 2000). Regardless of the direction, self-presence is related to the
concept presence and thus males and females are expected to experience the three
levels of self-presence differently.
Hypothesis 2a: Proto self-presence will differ between males and females.
Hypothesis 2b: Core self-presence will differ between males and females.
Hypothesis 2c: Extended self-presence will differ between males and females.
Avatar customization has been found to affect the extent to which people feel
connected to their avatars (Lim & Reeves, 2009) and should serve as another potential
validity test of the SPQ. Differences in customization can manifest in many ways,
from the amount of time spent customizing an avatar, to the number or frequency of
changes made to an avatar, to the choice of a default versus customized avatar.
Regardless of the means, customization affects the identity characteristics presented
by the avatar to others in the virtual environment and thus should be positively related
to extended self-presence.
Hypothesis 3: Measures of avatar customization will be positively related to
extended self-presence.

25
Chapter 2: Study 1 (Pilot)
As an early test of internal consistency and validity, the SPQ was administered
to a small group of virtual world users. Unfortunately, there were not enough
participants in this pilot study (N = 31) to conduct a reliable factor analysis. Still, the
results presented here set a precedent for examinations of larger datasets. Given this
caveat, the primary intent of this analysis was to assess whether the items of the SPQ
form distinct but interrelated factors, as the conceptual framework suggests. Further,
the SPQ’s construct validity was explored by examining whether extended self-
presence was related to avatar customization variables (Hypothesis 3), such as avatar
design time and appreciation of the virtual world experience.
Chapter 2: Method – Study 1 (Pilot)
Participants and Context
Participants included 31 students (24 male, 7 female) between the ages of 23
and 46 (M = 26.16, SD = 4.25) with different cultural and educational backgrounds
from two universities in Switzerland. The pilot study was carried out within an
interdisciplinary seminar on “Strategic Management and Artificial Intelligence”,
entitled “Managing the Unexpected”, in the spring of 2009. Students met physically
for the seminar throughout the semester, but teams of four to five students were
comprised of members from each university in order to encourage the use of
electronic media for collaboration on group assignments. Although using the 3D
virtual world for collaboration was not mandatory, they were required to write
individual papers on their experience with the 3D virtual environment and to
participate in two feedback sessions with their supervisors in a mixed reality setting
using a combination of the 3D virtual world and video-conferencing (see Figure 2.1).
Surveys were administered as online questionnaires after the second feedback session.

26
Figure 2.1. Seminar feedback session in a mixed-reality setting with video-
conferencing and a 3D virtual world.

Materials and Measures
A Colloborative Virtual Envirnoment CVE (called “UNIworld”) was offered
to the students for their virtual teamwork. UNIworld was designed based on Sun
Microsystems' Project Wonderland toolkit (version 0.4). It offered various
collaboration features, such as in-world video players, application sharing, and real-
time webcam streams of the physical rooms. The students communicated via text
chat, “immersive audio” (i.e., voice communication that accounts for distance
between avatars), and a set of basic avatar gestures and postures.
Freedom to customize avatar appearance was limited (as compared to similar
commercial applications) in this field study. Participants began with cartoonish
gender-neutral avatars (see Nigel, bea and Thierry in Figure 2.2) and were able to
customize gender (binary male/female choice), clothing accessories (e.g., hats or
glasses), and color of skin, eyes, hair, pants, shirt, and shoes. These customizations
were categorical and did not allow for gradual changes. All students were able to
name their avatars and provide profile information (real name, sex, birthday,
nationality, language, occupation, university, major, and minor) that was available to

27
other students. These limitations in avatar customization detract from the
generalizability of the results in this chapter.
Figure 2.2. International virtual team meeting in UNIworld, the 3D virtual
environment used for student collaboration.

An early formulation of the SPQ was used to assess self-presence (see Chapter
Appendix). It should be noted that the questions addressing extended self-presence
were changed considerably after this pilot study was conducted because the
formulation utilized here confounds extension of identity with extent of effort in
avatar creation. Social presence was measured with a scale established for use in
virtual reality contexts (Bailenson, Blascovich, Beall, & Loomis, 2001). Cronbach’s
alpha for this five-item measure was .75. Avatar design time was measured with the
question, “How much time did you spend designing and modifying your avatar
appearance in Wonderland?”

28
Chapter 2: Results – Study 1 (Pilot)
An exploratory principal-axis factor analysis with an oblique rotation, because
dimensions of self-presence are conceptualized to be interrelated (Kim & Mueller,
1978), provided a preliminary assessment of the SPQ. This analysis indicated that all
of the items formed distinct factors that were consistent with the proposed levels of
self-presence (proto, core, and extended), supporting Hypothesis 1. The pattern and
structure matrices can be found in Tables 2.1 and 2.2. Although the factor analysis
resulted in four factors, the third factor contained only one item with a loading over
0.40 in the pattern matrix and 0.55 in the structure matrix. This anomaly was most
likely due to the small sample size and thus the third factor is not addressed.
Cronbach’s alpha for the proto self-presence items was .94, for the core self-presence
items was .93, and for the extended self-presence items was .95, indicating high
internal consistency for each factor. Thus, the items were averaged in order to create
composite measures of each level of self-presence.
A zero-order Pearson correlation analysis was used to test the interrelations
between the levels of self-presence as well as avatar design time, while a series of
independent samples t-tests were used to examine the relationship between participant
sex and the three levels of presence. As the framework of self-presence suggests, the
levels of self-presence were positively related to each other, also supporting
Hypothesis 1. No significant differences were found between males and females with
respect to proto self-presence, t(29) = -.86, p = .39, core self-presence, t(29) = -.93, p
= .36, or extended self-presence, t(29) = -1.63, p = .15. Thus, Hypotheses 2a-c were
unsupported. Both proto and extended self-presence were positively related to self-
reported avatar design time, though the latter far more than the former, supporting
Hypothesis 3. The correlation coefficients for these relationships are in Table 2.3.

29
Table 2.1. Pattern Matrix of SPQ
Component
1 2 3 4
Proto1 .86 .00 -.16 .02
Proto2 .82 .10 -.19 .10
Proto3 .91 -.09 .28 -.01
Proto4 .49 .18 -.24 .41
Proto5 .57 .18 -.30 .33
Proto6 .62 -.25 .26 .41
Proto7 .85 .15 .08 -.04
Core1 .44 .00 -.22 .52
Core2 .32 .19 -.37 .58
Core3 -.14 -.02 .20 .97
Core4 .09 .03 .08 .85
Core5 .19 .21 -.05 .71
Extended1 -.08 .83 .28 -.04
Extended2 -.09 .88 .19 -.21
Extended3 .21 .54 .10 .14
Extended4 .02 .61 .04 .29
Extended5 .06 .83 .05 .13
Extended6 .29 .32 .36 .23
Extended7 .55 .41 .38 -.08
Extended8 .48 .46 .22 -.02
Extended9 -.05 .92 -.23 .18
Extended1 .05 .30 .73 .14
Extended1 .22 .87 -.03 -.11

Table 2.2. Structure Matrix of SPQ
Component
1 2 3 4
Proto1 .86 .31 -.11 .50
Proto2 .91 .41 -.10 .50
Proto3 .89 .38 .31 .4
Proto4 .78 .43 -.14 .74
Proto5 .81 .42 -.19 .70
Proto6 .76 .21 .23 .69
Proto7 .89 .51 .18 .49
Core1 .72 .27 -.18 .77
Core2 .71 .39 -.27 .82
Core3 .41 .28 .201 .89
Core4 .59 .35 .11 .91
Core5 .67 .48 .04 .88
Extended1 .25 .88 .54 .17
Extended2 .16 .83 .46 .01
Extended3 .51 .70 .29 .42
Extended4 .44 .72 .24 .49
Extended5 .48 .91 .32 .42
Extended6 .57 .61 .48 .50
Extended7 .69 .73 .54 .36
Extended8 .67 .72 .40 .39
Extended9 .42 .88 .07 .43
Extended10 .30 .60 .83 .28
Extended11 .52 .92 .27 .29

Table 2.3. Pearson correlations
Proto Core Extended
SPQ_Proto 1.00

SPQ_Core .80** 1.00
SPQ_Extended .58** .51** 1.00
Avatar_design_ti .37* .31 .76**
N = 31, Note: *p < .05, **p < .01

30
Chapter 2: Study 2
In order to provide more robust internal consistency and validity analyses of
the concept of self-presence and the SPQ, the same hypotheses proposed for the pilot
study were reexamined using a larger sample size. While similar to the pilot study,
there were some notable differences in methodology.
Chapter 2: Method – Study 2
Participants and Context
This study took place within the auspices of the ShanghAI Lectures, a global
virtual lecture series on “Embodied Intelligence”, in the fall of 2009. The lecture
series was presented by the Artificial Intelligence Lab of the University of Zurich
from Shanghai Jiao Tong University to a global audience involving 15 universities
from five continents, which implies greater generalizability for the results from this
study. Two hundred eighty-two students (223 men and 59 women) collaborated
remotely in three- to four-member international virtual teams on mandatory weekly
group assignments and projects over the course of the semester. Unlike in the pilot
study, participants in this study did not meet physically, and so the results are based
purely on virtual interaction.
Participation in the research project, i.e., using a 3D Collaborative Virtual
Environment (CVE) to collaborate on the group assignments, was an optional part of
the lectures which had no bearing on participants’ academic evaluation. About one
third of the students (N=81) used the CVE and filled in the questionnaires. The
participants were 59 men and 22 women between the ages of 20 and 40 (M = 24.73,
SD = 3.15). Because of the higher proportion of males in this sample, the comparison
of self-presence between genders (Hypotheses 2a-c) is used to address potential
concerns.

31
The online registration form contained detailed information about the purpose
of the research project, the in-world behaviors and questionnaire measures that would
be collected, and the privacy-protection procedures. Online registration also required
that participants read and respond to an informed consent form. Unlike the pilot
study, there were no video streams incorporated into this study. Surveys were
administered after the final group exercise at the end of the semester. An incentive
was introduced to increase the rate of participation in the online surveys. The students
who filled in all questionnaires were invited to take part in a competition to win a 100
Euro voucher for Amazon.com.
Materials and Measures
As in the pilot study, the CVE utilized in this study was called UNIworld.
Participants' behavior in the virtual world, including the number of avatar-appearance
changes as a measure of avatar customization (Hypothesis 3), was automatically
recorded, and all other measures were collected via survey. The SPQ was
administered with modifications to the measure of extended self-presence, as
described earlier. The version utilized in this study, which can be found in the
Chapter Appendix, does not vary as drastically throughout the rest of the dissertation.
Chapter 2: Results - Study 2
As in the pilot study, a principal-axis factor analysis with an oblique rotation,
because dimensions of self-presence are conceptualized as interrelated (Kim &
Mueller, 1978), provided an assessment of the SPQ’s factor loadings. This analysis
was confirmatory with respect to the three expected factors, but also exploratory in
that the factor loadings determined whether items were retained. After removing
items that loaded onto more than one factor or had low loadings, the remaining 16
items formed distinct factors that were mostly consistent with the proto, core, and

32
extended levels of self-presence. The eigenvalues for these factors were 7.34, 2.08,
and 1.35, respectively. The pattern and structure matrices can be found in Tables 2.4
and 2.5.
While these results lend partial support to Hypothesis 1, unexpectedly, some
of the extended self-presence items did not factor consistently with the others. A
further factor analysis, conducted only on the extended self-presence items, indicated
that there were two distinct extended self-presence factors, one with four items and
the other with three items. The pattern and structure matrices for this analysis can be
found in Tables 2.6 and 2.7.

33
Table 2.4. Pattern Matrix of SPQ

Factor
1 2 3
P_part_of_bo 0.93 -0.04 0.02
P_arm_elong 0.87 -0.10 0.02
P_reach_in 0.79 -0.08 -0.08
P_body_exte 0.78 -0.05 -0.06
C_disgusting -0.06 -0.07 -0.96
C_upsetting -0.10 0.05 -0.91
C_arousing 0.07 -0.01 -0.90
C_surprising 0.12 -0.01 -0.83
C_scary 0.08 -0.03 -0.82
E_gender -0.16 0.90 -0.02
E_sex 0.01 0.80 0.14
E_race -0.07 0.72 -0.12
E_clothing 0.38 0.42 -0.01
E_profile 0.47 0.33 -0.17
E_name 0.34 0.33 -0.13
E_appearance 0.44 0.29 -0.26

Table 2.6. Pattern Matrix of Ext SP
Factor
1 2
E_profile 0.93 0.09
E_appearance 0.83 0.06
E_name 0.63 -0.05
E_clothing 0.55 -0.17
E_gender -0.02 -0.91
E_sex -0.07 -0.85
E_race 0.16 -0.64

Table 2.5. Structure Matrix of SPQ

Factor
1 2 3
P_part_of_body 0.91 0.17 -0.55
P_arm_elong 0.84 0.10 -0.48
P_reach_in 0.82 0.13 -0.54
P_body_extend 0.81 0.16 -0.52
C_disgusting 0.61 0.28 -0.94
C_upsetting 0.51 0.21 -0.90
C_arousing 0.62 0.27 -0.90
C_surprising 0.46 0.30 -0.87
C_scary 0.57 0.24 -0.86
E_gender 0.07 0.87 -0.20
E_sex 0.11 0.76 -0.11
E_race 0.18 0.74 -0.30
E_clothing 0.48 0.52 -0.37
E_profile 0.65 0.50 -0.55
E_name 0.67 0.47 -0.62
E_appearance 0.49 0.44 -0.43

Table 2.7. Structure Matrix Ext SP
Factor
1 2
E_profile 0.89 -0.35
E_appearance 0.81 -0.34
E_name 0.65 -0.34
E_clothing 0.63 -0.43
E_gender 0.41 -0.89
E_sex 0.34 -0.82
E_race 0.46 -0.72

34
A reliability analysis was also conducted on each set of items. Cronbach’s
alpha for proto self-presence was .93, for core self-presence was .96, for all extended
self-presence items (now coined “Ext_All”) was .84, for the four-item extended self-
presence factor (now coined “Ext_Main”) was .82, and for the three-item extended
self-presence factor (now coined “Ext_Secondary”) was .84, indicating high internal
consistency for the items within each factor. Thus, the items were averaged in order to
create composite measures of each factor..
A zero-order Pearson correlation analysis was used to test the interrelations
between the levels of self-presence as well as number of avatar changes, while a
series of independent samples t-tests were used to examine the relationship between
participant sex and the three levels of presence. Most of the factors of self-presence
were positively related to each other. However, the secondary measure of extended
self-presence (Ext_Secondary) was not significantly correlated with any of the other
factors. Thus, Hypothesis 1 was only partially supported. No significant differences
were found between males and females with respect to proto self-presence, t(96) = -
1.08, p = .29, core self-presence, t(96) = -.79, p = .43, the full extended self-presence
measure (Ext_all), t(89) = -0.07, p = .95, the main extended self-presence measure
t(92) = -4.10, p = .68, or the secondary extended self-presence measure
(Ext_Secondary), t(95) = 0.42, p = .68. Thus, Hypotheses 2a-c were unsupported.
Unexpectedly, the number of avatar changes was not correlated with extended self-
presence. Thus, Hypothesis 3 was unsupported. The correlation coefficients for these
relationships can be found in Table 2.8.

35
Table 2.8. Pearson correlation between SPQ factors and other variables.
Proto Core E_All E_Main E_Sec
SPQ_Proto 1
SPQ_Core .63** 1
SPQ_E_All .41** .48** 1
SPQ_E_Main .59** .59** - 1
SPQ_E_Sec .06 .20 - .44** 1
#Avy Changes .01 -.05 .04 .00 .07
N = 81, Note: *p < .05, **p < .01

Chapter 2: Method - Study 3
This study provides an even more robust examination of the hypotheses tested
in the previous two studies. The methodology is nearly identical to that of Study 2,
with students across five continents from 20 universities participating within the
auspices of the ShanghAI Lecture series in the fall of 2010. For this iteration, data
was collected from 197 participants (151 males, 46 females) who collaborated in four
to five member virtual teams throughout the semester. In contrast to Study 2, in
addition to the group exercises, participants participated in bi-weekly discussion
sessions with their lecturer within the virtual environment. Further, participation in
the research project was mandatory and groups in which all members completed the
surveys received extra credit. These two differences from Study 2 induced a higher
response rate (80%), with 159 (123 males, 36 females) student respondents between
the ages of 19 and 49 (M = 23.60, SD = 3.33). The structure of the assignment and
virtual context were the same as in Study 2, though there was one additional avatar
customization option. Namely, participants had the option of uploading their own
picture onto their avatars’ faces using the “evolver” server (http://www.evolver.com,
Darwin Dimensions, Inc.,). This allowed for a more in-depth analysis of the
relationship between avatar customization and extended self-presence (Hypothesis 3).
Specifically, avatars customized with in-world tools represent a different type of

36
customization than avatars customized with the users’ face pictures. The former is
referred to customization and the latter as personalization. This study’s final
questionnaire included the same version of the SPQ as in Study 2 as well as a self-
report measure of the type of avatar used (default/uncustomized, customized using in-
world options, or customized with own picture). In order to determine technological
experience or ease, participants were asked to report the number of hours they spend
using a computer each week as well as how frequently they play video games and use
virtual worlds (5-point likert scales ranging from “never” to “every day”).
Chapter 2: Results - Study 3
As in the previous two studies, a principal-axis factor analysis with an oblique
rotation, because dimensions of self-presence are conceptualized as interrelated (Kim
& Mueller, 1978), provided an assessment of the SPQ’s factor loadings. This analysis
was confirmatory of the three expected factors, but also exploratory in that the factor
loadings determined whether items were retained.
The initial factor analysis indicated that some of the extended self-presence
items did not load strongly on this factor. A further factor analysis, conducted only
on the extended self-presence items, indicated that items regarding gender and sex
formed a separate factor (Tables 2.9 and 2.10). Although the distinction between the
two items is valid (i.e., someone could make a masculine female avatar or a feminine
male), their meaning is similar, which may have been confusing to some of the
participants. Thus the sex item was removed in a subsequent factor analysis of the
overall SPQ (Tables 2.11 and 2.12).

37
Table 2.9. Extended SP Pattern Matrix

Factor

1 2
E_clothing 0.84 -0.02
E_appearance 0.81 -0.08
E_profile 0.75 -0.06
E_race 0.66 0.14
E_name 0.39 0.17
E_gender 0.04 0.91
E_sex -0.01 0.74

Table 2.11. Final SPQ Pattern Matrix

Factor

1 2 3
C_surprising 0.89 0.06 0.07
C_happy 0.88 0.03 0.10
C_scary 0.82 -0.07 -0.08
C_arousing 0.81 0.06 -0.03
C_upsetting 0.80 0.03 -0.06
C_disgusting 0.78 -0.03 -0.12
C_sad 0.63 0.03 -0.20
E_race 0.01 0.84 0.06
E_clothing 0.09 0.65 -0.17
E_gender 0.06 0.61 0.14
E_appearance 0.18 0.57 -0.12
E_profile 0.02 0.44 -0.43
E_name -0.08 0.41 -0.23
P_handinside 0.01 -0.05 -0.89
P_partofbody 0.05 0.06 -0.84
P_extension 0.09 0.05 -0.72
P_arm_elong 0.22 0.00 -0.62
P_reach_in 0.24 0.06 -0.56
Table 2.10. Ext SP Structure Matrix

Factor

1 2
E_clothing 0.83 0.41
E_appearance 0.78 0.34
E_race 0.73 0.48
E_profile 0.72 0.32
E_name 0.47 0.37
E_gender 0.50 0.93
E_sex 0.37 0.73

Table 2.12. Final SPQ Structure Matrix

Factor

1 2 3
C_surprising 0.87 0.46 -0.52
C_arousing 0.86 0.46 -0.57
C_upsetting 0.85 0.44 -0.58
C_disgusting 0.84 0.40 -0.60
C_scary 0.83 0.36 -0.57
C_happy 0.82 0.41 -0.47
C_sad 0.77 0.42 -0.61
E_race 0.37 0.82 -0.32
E_clothing 0.51 0.77 -0.52
E_appearance 0.53 0.71 -0.48
E_profile 0.51 0.64 -0.64
E_gender 0.26 0.57 -0.16
E_name 0.26 0.47 -0.36
P_partofbody 0.61 0.45 -0.90
P_handinside 0.56 0.35 -0.88
P_extension 0.57 0.40 -0.79
P_arm_elong 0.61 0.38 -0.76
P_reach_in 0.63 0.42 -0.74

38
This version of the factor analysis provided factors that were consistent with
core, extended, and proto self-presence and had eigenvalues of 9.27, 1.78, and 1.42,
respectively. Thus, the individual items for each factor were averaged to form
composite measures of proto, core, and extended self-presence. The Cronbach’s
alphas for these three composite measures were .91, .94, and .83, respectively. A zero-
order Pearson correlation analysis (Table 2.13) indicated that the three levels of self-
presence were highly interrelated. Thus, Hypothesis 1 was supported.
It is worth noting that the structure matrix indicated somewhat high cross
loadings (over .5) for some of the factors. While the pattern matrix reflects causal
weights between an item and a factor, the structure matrix reflects correlations not
only between the items and the given factor but also between multiple factors and
individual items (Kim & Mueller, 1988). Thus, because the factors were strongly
correlated, the structure matrix indicated high cross loadings. To ensure that the three
factors were indeed distinct, discriminant validity was established (Fornell & Larcker,
1981) by utilizing path analysis software (Ringle, Wende, & Alexander, 2005).
Specifically, the average variance extracted (AVE) for the proto self-presence items
(0.91), core self-presence items (0.99), and extended self-presence items (0.88) were
compared to the square of the correlations between proto and core self-presence
(0.49), core and extended self-presence (0.31), and proto and extended self-presence
(0.31). The AVE values were all greater than the squared correlations, and so the
factors are indeed distinct (Fornell & Larcker, 1981).
Table 2.13. Pearson correlation between SPQ factors, and other variables.
Proto Core Extended
Proto 1
Core .70** 1
Ext .56** .55** 1

39

A series of independent samples t-tests were used to examine the relationship
between participant sex and the three levels of presence. No significant differences
were found between males and females with respect to proto self-presence, t(157) =
0.72, p = .47, core self-presence, t(157) = 0.16, p = .87, or extended self-presence,
t(157) = -0.15, p = .88. Thus, Hypotheses 2a-c were unsupported.
A series of Univariate Analysis of Variance (ANOVA) tests were conducted
in order to examine the relationship between the type of avatar used and the levels of
self-presence. The independent variable in each analysis, type of avatar, could take
three levels: default/uncustomized, customized using in-world options, or customized
with own picture. The dependent variable for each analysis was the respective level
of self-presence. The number of hours spent using a computer each week as well as
the frequency of playing video games and using virtual worlds were used as
covariates. The effect of avatar type was significant for proto self-presence, [F(2,
144) = 12.07, p < 0.001], core self-presence, [F(2, 144) = 13.15, p < 0.001], and
extended self-presence, [F(2, 144) = 10.85, p < 0.001].
Post hoc comparisons using the Tukey HSD test indicated that proto self-
presence was significantly higher for the customized-with-own-picture condition
compared to both of the other conditions (p < .01), but that there was no difference
between the default/uncustomized condition and the customized using in-world
options condition. The post hoc test produced the same pattern of results for core
self-presence. A slightly different pattern was found for extended self-presence.
Namely, extended self-presence was significantly lower in the default/uncustomized
condition than both the customized using in-world options condition (p < .05) and the
customized with own picture condition (p < .01), but there was no significant

40
difference between these latter two conditions. This finding regarding extended self-
presence supports Hypothesis 3. Namely, using a customized or personalized avatar
facilitates extended self-presence. However, it appears that proto and core self-
presence are related only to personalized, but not customized avatar use. A table with
means and standard deviations that reflect these tests can be found in Table 2.14.
Table 2.14. SPQ descriptives by avatar type
Avatar Type Mean Std. Dev N
Proto Default/Uncustomized 2.19 .96 71

Customized w in-world options
2.14 0.89 33

Customized with own picture
2.96 0.70 40
Core Default/Uncustomized 2.27 1.10 71

Customized w in-world options
2.19 0.86 33

Customized with own picture
3.06 0.61 40
Extended Default/Uncustomized 2.41 0.97 71
Customized w in-world options
3.03 0.82 33

Customized with own picture
2.76 0.77 40

Chapter 2: Discussion
Results from three studies support three basic hypotheses that are fundamental
to establishing the internal consistency and validity of self-presence and the Self-
Presence Questionnaire (SPQ) measurement tool. All three studies found that the
SPQ factors were mostly internally consistent and that these factors were positively
interrelated, which supports the conceptual framework of self-presence. Further,
Studies 1 and 3 found that avatar customization is positively related to extended self-
presence, supporting the validity of this level of self-presence. Finally, Study 3 found
a relationship between avatar personalization and proto and core self-presence,
suggesting that an important distinction between avatar customization and
personalization.

41
While the factor analyses in all three studies yielded results that were mostly
consistent with the framework of self-presence, unexpectedly, in Studies 2 and 3,
some items formed a subfactor of extended self-presence. Specifically, in Study 2,
the items relating to race, gender, and sex formed their own factor, while in Study 3,
the gender and sex items formed their own factor. Regarding Study 2, the likely
explanation for this finding is that the type of customization allowed by the virtual
environment may have favored the expression of some identity characteristics over
others. Namely, participants were given more freedom to customize facets of avatar
appearance that related more to items in the main extended self-presence factor than
in the secondary factor. For example, they could create any name for their avatars,
but could only make a binary choice of avatar gender (male or female). Similarly,
there were many possible combinations of colors for clothing items, and while avatar
skin and eye color options were customizable, there are many more race-related facets
of appearance that can be expressed in an avatar (Groom, Bailenson, et al., 2009). In
Study 3, the race item did not load onto the secondary subfactor of extended self-
presence, but instead factored strongly with the main extended self-presence factor,
possibly because some participants used personalized avatars with pictures of their
own faces. Because the two items in the subfactor, gender and sex, have similar
meanings, the latter was removed from the analysis. The subsequent extended self-
presence factor had higher internal consistency, with all items in the pattern matrix
loading over 0.55.
While the concept of self-presence is applicable across all virtual
environments, this series of tests indicates that the SPQ should be treated as an
inventory of potential self-presence questions from which researchers should select
appropriately depending on the specific customizable facets of avatars in the

42
environment. For example, if a virtual environment facilitates a high level
customization of both gender and sex avatar attributes (i.e., users can make masculine
female avatars or feminine males), such as in the social virtual world Second Life,
then the researcher should include both of these items. Otherwise, only the gender
item should be included. In many cases, it may be difficult to anticipate which items
are appropriate, and so researchers should include as many items as possible and
conduct their own factor analyses on the SPQ to identify which items to retain.
Researchers may choose to run a factor analysis on the entire SPQ or separate
analyses on the items for each level of self-presence. As the present findings suggest,
if the latter method is employed, the researcher should pay particular attention to the
extended self-presence items. Regardless of method, the researcher should report
which items are included in the final analysis and provide possible explanations for
why any items were removed.
Although males and females have been found to experience presence
differently, no difference was found between males and females for self-presence in
any of the three studies. This may indicate that self-presence is distinct from presence
in this regard. However, such inferences from a non-finding should be treated
cautiously. The reason for this lack of a difference between males and females may
stem from the specific type of medium and context, a collaborative virtual
environment being used for a school-related purpose. Future research should examine
this issue further by testing for differences in self-presence between males in females
in additional types of virtual environments.
The findings related to avatar customization also provide validity support as
well as notable implications. The strong relationship between extended self-presence
and avatar design time in Study 1 makes sense because the measure of extended self-

43
presence in this study included questions gauging the extent of effort put into creating
the avatar. But given that this measure changes in future iterations of the SPQ and the
small sample size, the support for validity is limited. In Study 2, contrary to
expectations, extended self-presence was not related to the number of avatar
appearance changes. While this seems surprising initially, the finding makes sense
when considering the difference between avatar customization and personalization.
The former does not necessarily contribute to the latter. Participants may have made
many changes to their avatars, but these changes may not have been targeted at
imbuing the avatar with characteristics that reflect their personalities. Perhaps
participants were not motivated to personalize their avatars because of the limited
options for customization or some attribute of the course assignment. Regardless of
the reason, this finding, or lack thereof, raises the important point that customization
does not always lead to personalization and thus that the latter may be a more
important contributor to extended self-presence than the former.
The findings related to avatar selection in Study 3 help to elucidate this
issue, provide validity support for extended self-presence, and raise a new point about
the relationship between avatar personalization and both proto and core self-presence.
Of the three avatar options in this study – 1) default/uncustomized, 2) customized
with in-world options, and 3) customized with own picture – the latter two reflect a
distinction between avatar customization and personalization. Namely, if a
participant customizes an avatar with in-world options, this does not necessarily imply
that this avatar will be personalized, as just inferred from Study 2. However, if a
participant customizes an avatar with her own picture, this avatar will definitely
reflect her personal identity to some extent. While it is conceivable that an avatar
could be made more personalized using in-world options to customize various non-

44
facial attribute (e.g., body size, clothing), given the limitations of customization in the
present virtual environment, this is unlikely. Still, this does not necessarily mean that
avatars customized with in-world options will be less personalized than those with
own-face pictures. The relationship between avatar type and extended self-presence
reflects this. Specifically, there was no significant difference between the two types
of customized avatars, but both types were associated with significantly more
extended self-presence than the default/uncustomized avatar, which supports the
construct validity of this level of self-presence. But more interesting, perhaps, is the
distinction between avatar types with respect to proto and core self-presence.
Namely, while there were no differences between the default/uncustomized avatar
type and the customized with in-world attributes avatar type, the own-picture
customized avatar was rated significantly higher for both proto and core self-presence.
Given that the previous finding implies that there should be no difference between the
two types of customized avatars in terms of extent of personalization, there must be
some other mechanism driving this difference. Perhaps using an own-face avatar
reduces the feeling of anonymity, thereby creating a different type of personalization
in the avatar that facilitates the feeling of body and emotion-level connection to the
avatar. Alternatively, a trait-based interpretation is that technological ability or
motivation may influence both the type of avatar that people choose as well as the
extent of body and emotion-level connections to their avatars. Given that the
variables of time spent on computers per week and frequency spent playing video
games and using virtual worlds were used as covariates in this test, this latter
explanation seems unlikely. Still, there may be other state-based interpretations worth
considering. The question of state versus trait-based interpretations arises
consistently throughout this dissertation, thus indicating that future studies about or

45
using self-presence should utilize designs that control for these issues (e.g., pre-test,
post-test).
Overall, the studies presented in this chapter represent an important first step
in establishing the construct’s internal consistency and validity, but their results are
not generalizable beyond the type of collaborative course-related virtual context that
they all utilized. Thus, an important next step for this dissertation is to examine the
SPQ within other types of virtual contexts.

46

Chapter 3: Self-Presence and Online Gamers

Chapter 3: Abstract
The present chapter examines responses from online gamers (N=307) to test
the internal consistency and validity of self-presence and the Self-Presence
Questionnaire (SPQ). Factor analysis yields results that are consistent with the three
proposed levels of self-presence, thus supporting the conceptual framework’s
reliability. Validity tests illustrate differences between participant sex and game
genres that are consistent with expectations. Skew in population sample reduces the
external validity of these findings, but internal validity is maintained. Overall, these
results provide support for the claim that self-presence is a valuable concept for
research on virtual self-representations and that the SPQ is an appropriate
measurement tool for this concept across various virtual contexts. It should be noted
that much of the content in this chapter comes directly from an existing paper (Ratan,
2010).

The present chapter provides support for the internal consistency and validity
of the concept of self-presence and the Self-Presence Questionnaire (SPQ) in online
games. Millions of people play these games regularly (Woodcock, 2008), unlike the
collaborative virtual environments examined in the previous chapter, and thus the
present examination provides a higher degree of generalizability. The expectations
with respect to self-presence are largely the same as in the previous chapter. First and
foremost, responses to the SPQ are expected to factor consistently with the framework
of self-presence, and these factors should be interrelated.

47

Hypothesis 1: Factor analysis of the SPQ will indicate that the levels of self-
presence (proto, core, and extended) are distinct but interrelated.
As described in the previous chapter, previous research has found that males
and females experience presence differently. The direction of this difference is not
consistent, with males reporting more presence than females in some cases (Lachlan
& Krcmar, 2011) and less presence than females in others (Lombard, et al., 2000;
Nah, et al., 2010). The similarity between presence and self-presence suggests that
there should also be sex differences for the latter. Although the previous chapter’s
studies did not find any, the current chapter focuses on a different type of virtual
environment in which such differences may occur. Thus, this issue is examined in a
similar manner as in the last chapter, without an expectation of the direction of the
relationship.
Hypothesis 2a: Proto self-presence will differ between males and females.
Hypothesis 2b: Core self-presence will differ between males and females.
Hypothesis 2c: Extended self-presence will differ between males and females.
Although many online games share similarities, such as social interaction,
there are large differences between the types of games that affect the players’
experiences in the game and thus their experiences of self-presence. Some of these
differences can be characterized systematically according to game genre, and so just
as differences in game type have been found to systematically affect players’
economic behaviors (Castronova, et al., 2009), there may be systematic differences in
self-presence across game genres . Some genres create fast-paced, first-person virtual
environments that are likely to induce proto self-presence, whereas others require
more social interaction and exchange of identity-relevant information, and thus are
more likely to induce extended self-presence. First-Person Shooters (FPS), in which

48

players navigate 3D environments from the first-person perspective and attempt to
shoot targets, qualify within the former category (Tamborini & Skalski, 2006), while
Massively Multiplayer Online Games (MMOs), in which players take a third-person
or bird’s-eye perspective within fantasy-based environments and complete tasks to
accumulate virtual resources, qualify within the latter (Williams, et al., 2006). Thus,
FPS players should report more proto self-presence than MMO players, while MMO
players should report more extended self-presence than FPS players. Another
difference between these genres is the range of emotions they induce in players.
MMOs often take place in fantasy worlds, contain rich narratives, and offer a variety
of activities in which players engage, such as questing, crafting, and socializing
(Williams, et al., 2008). In contrast, FPS games contain mostly action sequences
comprised of moving and shooting (Tamborini & Skalski, 2006). The former provide
the opportunity for players to experience a wider range of emotions than the latter and
thus are more likely to induce core self-presence. Together, the above reasoning leads
to the following hypotheses.
H3: FPS games induce more proto self-presence than MMO games.
H4: MMO games induce more extended self-presence than FPS games.
H5: MMO games induce more core self-presence than FPS games.
The examinations of potential differences in self-presence caused by
participant sex and game genre, as described above, are based on measures of the
three levels of self-presence taken from participants of both sexes and games within
both genres of interest. This approach does not account for potential differences in
the reliability of such composite measures of self-presence between the participant
sexes and game genres. For example, it is possible that the SPQ is a reliable measure
of self-presence for males but not females, or for MMO players but not FPS players.

49

The examination of such potential differences is articulated through the following
research questions:
RQ1: Is the SPQ reliable for both males and females?
RQ2: Is the SPQ reliable for both MMO and FPS players?
Chapter 3: Method
The SPQ was administered as part of an online survey, which was distributed
via convenience sampling to email lists dedicated to video game and virtual world
research, game-related web forums, and individual players. The solicitation requested
that online gamers participate in a 10-minute survey about how people use and view
their avatars. The online survey tool restricted responses to one per IP address,
though no incentive for participation was provided. A link in the message took
respondents to an information and consent page, and if they agreed, another link took
them to the survey. Before answering the questions, participants were prompted to
consider the game they play online (“with others”) most often. The genre of the
games were coded by an online game researcher (see Chapter Appendix) as 1) MMO,
2) FPS, or 3) Other. It should be noted that social virtual worlds, such as Second Life
and Active Worlds, were coded as “Other” because they do not include structured
goal-directed gaming elements and thus the uses and motivations of such worlds have
been found to differ from more traditional MMOs (Ducheneaut, Wen, Yee, &
Wadley, 2009). Further, 96% of the MMO players play the game World of Warcraft,
which is the most widely played MMO (Woodcock, 2008) and thus representative of
the genre. Of the 337 respondents, only those who play an MMO or FPS game were
retained for the analysis, and thus 30 were removed. This resulted in a sample of 307
participants (218 female and 89 male) between the ages of 18 and 62 (m = 26.55, SD
= 6.92).

50

The distribution of participants was unbalanced for game genre and not
representative of online gamers according to participant sex. Either the distribution
channels for the survey reached more MMO players or MMO players were more
likely to respond to the survey. Regardless of the reason, only 28 participants (9%) in
the sample were FPS players. Further, a disproportionately high number of females
responded to the survey, perhaps also due to the distribution channels or to the nature
of the inquiry. Females currently make up 42% of online gamers (ESA, 2010), but in
the current sample 217 of 307 (71%) were female. In the MMO-player sample, 215
(77%) were female, which is the reverse of previous findings that less than 20% of
MMO players are female (Williams, Yee, & Caplan, 2008; Yee, 2006). Levene’s test
for equality of variances can be used to account for this discrepancy in sex
distribution within MMO players, but results must still be treated cautiously. Further,
within the FPS-player sample, only 2 (7%) were female, which is consistent with
previous FPS research (Jansz & Tanis, 2007) but inhibits the ability to draw statistical
inferences about participant sex differences within this genre.
Overall, these skews confound the tests of sex and genre differences for self-
presence. In other words, when considering the entire sample, changes in sex are
always accompanied by large changes in genre, and vice versa. The solution to this
problem adopted here, which sacrifices external for internal validity, is to limit the
sample population within which the tests are conducted. Specifically, the tests for sex
differences, Hypotheses 2a-c, are conducted only for MMO players while the tests for
genre differences, Hypotheses 3-5, are conducted only for males.
Chapter 3: Results
Because the dimensions of self-presence are conceptualized to be interrelated,
a principal-axis factor analysis with an oblique rotation (Kim & Mueller, 1978) was

51

used to determine whether responses to the SPQ aligned with the framework of self-
presence. As in the previous chapter’s studies, although this analysis was designed to
confirm expected factors, it was exploratory in that the factor loadings determined
whether items were retained or removed. Specifically, items were retained only if the
loadings were greater than 0.50 on the target factors (with the exception of the
retained name-related extended self-presence question for which the structure matrix
loading was 0.49) and cross loading between factors were less than or equal to 0.45.
The remaining 18 items formed three distinct factors that were consistent with the
proposed proto, core and extended levels of self-presence and had eigenvalues of
2.36, 6.06 and 1.62, respectively. The items retained for the analysis are specified in
the Chapter Appendix.
The pattern and structure matrices can be found in Tables 3.1 and 3.2.
Cronbach’s alpha for proto self-presence was .88, for core self-presence was .85, and
for extended self-presence was .78, indicating high internal consistency for the items
within each factor. Thus, the items were averaged in order to create composite
measures of each level of self-presence.
A zero-order Pearson correlation analysis was conducted with the three
composite measures of each level within the SPQ in order to determine whether these
factors were interrelated. The correlations, which can be found in Table 3.3, though
not extremely strong, indicate that the levels are interrelated as well as distinct, and so
H1 was supported.
A series of independent samples t-tests was used to identify differences in self-
presence according to participant sex and game genre. As described earlier, the
former was examined only for MMO players and the latter only for males. Compared
to male MMO players, female MMO players reported significantly more proto self-

52

presence, t(120) = 3.00, p < .01, core self-presence, t(273) = 3.53, p < .001, and
extended self-presence, t(273) = 2.94, p < .01. Thus, Hypotheses 2a-c were supported
in the direction that females experience more self-presence than males. Descriptive
statistics for these tests can be found in Table 3.4. It should be noted that Levene’s
test for equality of variances was satisfied for core and extended self-presence, but not
for proto self-presence, which is why the degrees of freedom are reduced for the proto
self-presence t-test. Compared to male FPS players, male MMO players reported
significantly less proto self-presence, t(35) = -3.07, p < .01. Thus, Hypothesis 3 was
supported. Again, this t-test did not assume equal variances, which explains the
reduction in degrees of freedom. No differences were found between male FPS
players and male MMO players for core self-presence, t(85) = -.40, p = .69, or
extended self-presence, t(85) = .03, p < .98. The descriptive statistics for these game
genre comparisons can be found in Table 3.5.
RQs 1 and 2 address the issue of whether there are differences in the reliability
of the three self-presence composite measures according to participant sex or game
genre, respectively. A series of reliability analyses were conducted on the composite
measures of self-presence with the sample population segmented by participant sex
and then by game genre. The results, which can be found in Table 3.6, indicate that
all composite measures have acceptable Cronbach’s alphas (above .74) for all
segments.

53
Table 3.1. Pattern Matrix

Factor
1 2 3
P_arm_elongated
-0.10 -0.86 0.00
P_hand_inside_game
-0.10 -0.83 0.04
P_part_of_body
0.03 -0.83 -0.02
P_extension_of_body
0.03 -0.67 0.02
P_reach_into_game
0.14 -0.62 0.04
P_happens_to_body
0.11 -0.58 0.00
C_sad
0.80 0.09 0.03
C_upsetting
0.73 0.05 -0.01
C_suprising
0.69 -0.04 0.00
C_disgusting
0.65 -0.11 -0.06
C_scary
0.62 0.03 0.09
C_happy
0.60 -0.12 0.06
E_appearance -0.02 0.01 0.81
E_profile 0.06 -0.07 0.71
E_race -0.07 -0.01 0.70
E_clothing 0.08 0.07 0.64
E_name 0.18 -0.16 0.60

Table 3.2. Structure Matrix

Factor
1 2 3
P_part_of_body
0.34 -0.84 0.34
P_arm_elongated
0.24 -0.82 0.30
P_hand_inside_game
0.24 -0.81 0.33
P_reach_into_game
0.41 -0.70 0.37
P_extension_of_body
0.29 -0.68 0.30
P_happens_to_body
0.34 -0.62 0.30
C_sad
0.78 -0.23 0.40
C_suprising
0.71 -0.31 0.38
C_upsetting
0.70 -0.23 0.35
C_happy
0.68 -0.38 0.42
C_disgusting
0.67 -0.34 0.33
C_scary
0.66 -0.25 0.40
E_appearance
0.39 -0.33 0.80
E_profile
0.45 -0.39 0.76
E_race
0.39 -0.23 0.64
E_clothing
0.28 -0.24 0.57
E_name
0.24 -0.21 0.49

54
Table 3.3. Pearson correlations between self-presence levels
Proto Core Ext
Proto 1.00
Core
0.40*
* 1.00
Ext
0.38*
* 0.46** 1.00
** significant at the 0.01 level N = 307

Table 3.4. Descriptive statistics for MMO players by sex

Sex N Mean
Std.
Deviation
Male 62 1.53 0.67 Proto
Fem 213 1.83 0.82
Male 62 2.33 0.85 Core
Fem 213 2.72 0.73
Male 62 2.01 0.83 Ext
Fem 213 2.37 0.86

Table 3.5. Descriptive statistics for male players separated by game genre

Sex N Mean
Std.
Deviation
MMO 62 1.53 0.67 Proto Self-Presence
FPS 25 2.15 0.92
MMO 62 2.33 0.85 Core Self-Presence
FPS 25 2.41 0.75
MMO 62 2.01 0.83 Extended Self-Presence
FPS 25 2.00 0.87

Table 3.6. Cronbach's alphas for self-presence measures segmented by participant
sex and game genre

Males Females MMO FPS
Proto Self-Presence 0.87 0.88 0.88 0.85
Core Self-Presence 0.87 0.84 0.86 0.78
Extended Self-Presence
0.79 0.77 0.78 0.74

55
Chapter 3: Discussion
The present chapter provides support for the internal consistency and validity
of the concept of self-presence and the Self-Presence Questionnaire (SPQ)
measurement tool in the context of online games, implying greater generalizability of
the concept than the previous chapter offered. Responses to the SPQ aligned
distinctly with the proto, core, and extended levels of self-presence and these factors
were positively interrelated, supporting the framework’s internal consistency.
Females reported more proto, core, and extended self-presence than males, which is
consistent with previous research on the related topic of presence. Male FPS players
reported more proto self-presence than male MMO players, which supports the
construct’s validity. The lack of differences in core and extended self-presence
between male MMO and FPS players cannot be properly interpreted until future
research examines such a distinction for females.
Although consistent with the findings from the previous chapter, the present
support for the concept’s reliability is notable because it shows that the SPQ is
applicable to virtual contexts that millions of people choose to use every day
(Woodcock, 2008). While online games are designed uniquely and vary widely, these
findings indicate that the levels of self-presence are manifested similarly in MMO and
FPS games. Combined with the results from the previous chapter, self-presence
appears to be consistently reliable across virtual environments and user types. Still,
future research should continue to expand the reliability of self-presence by
examining it in new virtual contexts and with new populations.
The finding that female MMO players reported more self-presence across all
three levels than male MMO players may suggest that females tend to experience
more self-presence. However, such a conclusion should be treated cautiously because

56
the direction of difference between males and females with respect to presence has
been found to depend on various factors during media use (Lachlan & Krcmar, 2011;
Lombard, et al., 2000; Nah, et al., 2010). Further, the previous chapter’s three studies
found no such difference for participants in a collaborative virtual environment. The
most appropriate conclusion from this current finding is that self-presence may differ
according to participant sex and thus this should be taken into consideration when
examining self-presence in the future.
The finding that male FPS players reported more proto self-presence than
male MMO players supports the construct’s validity because FPS games contain more
fast-paced realistic sequences than MMOs and thus are expected to induce greater
proto self-presence. Although this difference could only be tested for males in the
present study, there is no reason to expect a different outcome for females. However,
the lack of differences in core and extended self-presence between male MMO and
FPS players may suggest that such differences do not occur at all or only occur for
females. The reasoning that MMOs should induce more core self-presence because
these games offer a greater diversity of emotional experiences may be flawed in
general because FPS games may offer a lesser diversity but greater intensity of
emotional experiences than MMOs. The reasoning that MMOs should induce more
extended self-presence because avatar customization is more integral to such games
may be flawed for males because FPS avatars, though limited in customization, are
often male and thus easier for males to relate to with respect to identity
characteristics. Thus, future research should examine genre differences with respect
to extended self-presence for females.
Although the reasoning for the expectations of genre differences in the three
levels of self-presence implies that characteristics of the game affect the experience of

57
self-presence, it is possible that that people who tend to enjoy experiencing certain
types of self-presence also choose to play games that facilitate these types of self-
presence. For example, some people may enjoy the feeling of proto self-presence and
thus choose to play FPS games more often than MMOs. This issue of trait-related
differences in the experience of self-presence arises in Chapter 7 and should be
examined more thoroughly in future research.
Overall, the results presented in this chapter contribute support to the claim
that the concept of self-presence and the SPQ measurement tool are applicable across
various virtual contexts. However, because these findings and those from the
previous chapter are cross-sectional, there are still open questions about what exactly
causes self-presence. In order to address this need, a study must incorporate a level of
control and manipulation so that the levels of self-presence can be examined
independently as causes and effects. The following chapter presents such a study.

58
Chapter 4: Controlled Manipulations of Self-Presence via Avatar Assignment

Chapter 4: Abstract
The present chapter examines self-presence within a controlled experimental
context. In the study, participants (N = 64) played a single-player sword-fighting
game using either a customized or generic avatar that was either similarly or
oppositely gendered with the participant. Results of a factor analysis of the SPQ are
consistent with those presented in the previous chapters, with responses to the SPQ
factoring according to the conceptual framework. Discriminant validity was
established between the three levels of self-presence and the concept of presence.
Supporting validity, extended self-presence was influenced by avatar customization,
as expected, but not by avatar gender consistency. Further, both proto and core self-
presence were influenced by avatar gender consistency but not customization,
indicating that these levels of self-presence may relate more to the physical attributes
of an avatar than to the act of adjusting those attributes. Finally, there was an
interaction effect between the avatar assignment conditions, which indicates that
using a generic, similarly gendered avatar or a customized, oppositely gendered avatar
detracts from the player’s feeling of physical connection with the avatar and thus
hinders proto self-presence. Overall, these results indicate that the SPQ is a reliable
and valid tool for use in this type of virtual context and further that the levels of self-
presence are influenced by avatar attributes in predictable ways. It should be noted
that much of the content in this chapter comes directly from an existing paper (Ratan,
2011).

59
The present chapter examines self-presence within a different type of virtual
environment from the previous chapters. Namely, the study presented here applies
self-presence to console-based video games, which differ in use habits from online
games and virtual worlds (ESA, 2010), but are also a popular type of environment in
which people create virtual self-representations. This study offers experiment-based
internal consistency and construct validity confirmations of the Self-Presence
Questionnaire (SPQ). The expectations regarding internal consistency are the same as
in the previous chapters, namely, that the SPQ should factor consistently with the
framework of self-presence.
Hypothesis 1: Factor analysis of the SPQ will indicate that the levels of self-presence
(proto, core, and extended) are distinct.
When developing a new concept, it is important to establish that the concept is
distinct from previously existing concepts, i.e., discriminant validity. As described in
Chapter 1, self-presence shares origins with and is similar to the concept of presence,
but the studies in the previous chapters were unable to reliably measure presence
because the context of avatar use was not controlled (outside of the virtual
environment). The present chapter offers a controlled examination of self-presence
and thus the opportunity to test for discriminant validity between presence and the
three levels of self-presence. As the conceptual framework suggests, each level of
self-presence is expected to be distinct from presence.
Hypothesis 2a: Proto self-presence is distinct from presence.
Hypothesis 2b: Core self-presence is distinct from presence.
Hypothesis 2c: Extended self-presence is distinct from presence.
In order for a controlled experiment to provide stronger support for the
construct validity of self-presence, responses to the SPQ should differ according to a

60
manipulated variable that has been found to affect a construct that is similar to self-
presence. One such variable is avatar customizability. This variable is externally
valid because the extent of customizability varies between specific virtual worlds and
video games, with some allowing detailed personalization of avatar appearance (e.g.,
Second Life; Robbins & Bell, 2008), and others offering choices of only pre-set
characters (e.g., Super Mario Brothers; Sherman, 1997). Previous research has found
that avatar customizability (choice versus assignment of an avatar) affects
physiological arousal during avatar use (Lim & Reeves, 2009). Specifically,
participants who chose their avatar reported higher arousal, indicating greater feelings
of ownership over their avatars. Thus, this implies that a person who is allowed great
avatar customization will experience higher levels of extended self-presence than one
who is restricted from avatar customization.
Hypothesis 3: Using a customized avatar will induce more extended self-presence
than using a generic avatar.
Gender is another important characteristic of avatars that is salient in many
virtual worlds and video games. Most people choose character gender to be
consistent with their own biological sex, presumably for identity consistency reasons
(Huh & Williams, 2010; Hussain & Griffiths, 2008). While some people who use
oppositely gendered avatars (“gender-jumpers”) do so for identity-driven reasons
(Roberts & Parks, 1999), most gender jumpers are motivated by functional reasons,
such as females using male avatars to avoid negative prejudice (Roberts & Parks,
1999; Yee, 2008), or males using female avatars to receive attention and gifts from
other players (MacCallum-Stewart, 2008). This implies that, overall, people who use
avatars that are similarly gendered with their sex will feel a stronger identity
connection, and thus extended self-presence, with that avatar.

61
Hypothesis 4: Using a similarly gendered avatar will induce more extended self-
presence than using an oppositely gendered avatar.
In Chapter 2 of this dissertation, the findings regarding avatar type in Study 3
indicated that participants who used a highly personalized avatar (customized with
participant’s own-face picture) reported higher levels of proto and core self-presence
than participants who used either a default/uncustomized avatar or an avatar that was
customized using in-world features only. This finding suggested an interesting
difference between avatar customization and personalization because only the latter
increased proto and core self-presence. However, the finding was confounded with
the act of choosing the specific avatar type, and so the present study aims to isolate
this effect through assignment to avatar type. The expected relationship between
avatar type and these levels of self-presence is consistent with the previous finding,
namely, that using a customized or similarly gendered avatar increases proto and core
self-presence, as expressed in the following set of hypotheses.
Hypothesis 5: Using a customized avatar will induce more proto self-presence than
using a generic avatar.
Hypothesis 6: Using a customized avatar will induce more core self-presence than
using a generic avatar.
Hypothesis 7: Using a similarly gendered avatar will induce more proto self-presence
than using an oppositely gendered avatar.
Hypothesis 8: Using a similarly gendered avatar will induce more core self-presence
than using an oppositely gendered avatar.
Finally, the interaction between avatar customization and gender consistency
warrants consideration. Though there is not much precedent, there may be systematic
ways in which using a customized, similarly gendered or generic, oppositely gendered

62
avatar differs from using a generic, similarly gendered or customized, oppositely
gendered avatar. While customization is expected to increase all three levels of self-
presence, customization of an oppositely gendered avatar may hinder self-presence
because people prefer to use similarly gendered avatars to maintain identity
consistency (Huh & Williams, 2010; Hussain & Griffiths, 2008). Similarly, while
gender consistency is expected to increase the levels of self-presence, using a generic,
similarly gendered avatar may reduce self-presence because the user would prefer to
be able to customize the avatar. Because neither the causal mechanism in this
relationship nor the expected differences between the levels of self-presence are
articulated in detail, the interaction is proposed in a single research question.
Research Question 1: Is there an interaction between avatar customization and gender
consistency with respect to the three levels of self-presence such that using a
customized, similarly gendered or generic, oppositely gendered avatar will induce
higher amounts of any of the levels of self-presence compared to using a generic,
similarly gendered or customized, oppositely gendered avatar.
Chapter 4: Method
Design and population
A 2 x 2 between-subject experiment was conducted at a large western
university to examine the hypotheses. The participant pool consisted of 64 right-
handed
2
female
3
undergraduate students between the ages of 18 and 28 (m = 19.83,
SD =2.07). Participants were randomly assigned to condition, with 16 participants in
each condition. The first independent variable was avatar customizability. Half of
the participants used an avatar that they had customized to look like themselves. Half

2
The sample was restricted to right-handed participants because the physiological recording equipment
(not discussed in this chapter) could only be attached to the left arm due to constraints in the lab.
3
The sample was restricted to females because of the stereotype threat induction described in Chapter
7.

63
were assigned to use a generic avatar. The second independent variable was avatar
gender. Half of the participants used a similarly gendered avatar. Half used an
oppositely gendered avatar. Figure 4.1 depicts the 2 x 2 for these four conditions
using a fictitious example of Robeanne as the participant and Roberto as the male
version of herself.
Figure 4.1. Avatar assignment examples for the experimental conditions.

In order to mitigate potential effects of using a specific avatar in the generic
condition, eight avatars were designed by the research team and then pretested by 104
convenience-sampled online respondents. The four (two male, two female) that were
found to be most average-looking were selected. Participants in the generic avatar
condition were randomly assigned to one of the two appropriately gendered avatars.
Materials
Wii Sports Resort, a suite of sports games for the Nintendo Wii, was played on
a Sony XBR 52” television. The specific game used within Wii Sports Resort was
Swordplay, a cartoon sword dueling game in which the player swings the controller
(“Wii mote”) to control a blunt ater sword and to knock an opponent off a platform.
Although the premise of the game may seem violent, the avatars are depicted wearing
safety gear and the game is rated by the Entertainment Software Rating Board
(ESRB) as appropriate for everyone. This game was chosen because the natural

64
interface was likely to induce variance in proto self-presence. Further, the camera in
the game is placed just slightly behind the avatar’s head, providing a hybrid first-
person/third-person perspective. The first-person is common in many presence-
inducing virtual environments (Biocca, 1997; Lee, 2004; Yee & Bailenson, 2007),
while the third-person perspective allows the player to see the avatar during play,
which is useful as a reminder of the avatar assignment. Other reminders of this exist
in the game, such as when players watch the replays at the end of the rounds or see
their avatars as icons during menu navigation.
Participants completed all surveys on an iPad while sitting in the same seat
used during game play (with the television off). This allowed for quick transitions
between experiment segments and also an easy method of entry for the participants.
This was useful because participants were hooked up to wired sensors for
physiological recording and could not easily change location within the room. The
data collected from this physiological equipment are not examined in the present
chapter, and so the equipment is not described here.
Procedure
Participants entered the lab and were instructed to sit on a couch and read the
IRB-approved study information sheet. After indicating to the researcher that they
had finished reading, they were given the iPad with a short demographic survey. The
researcher then hooked up the physiological equipment, which entailed placing or
asking the participants to place electrodes on the participants’ arms, finger, cheeks,
foreheads, and ankles. Participants were then instructed to spend five minutes
creating an avatar, or “Mii”, as it is called on the Wii. If the participant was in the
same-gender condition, she was told to make this Mii as similar to herself as possible.
If the participant was in the opposite gender condition, she was told to build the Mii

65
as a male version of herself. For additional clarification, she was told that this Mii
should appear to be how she would if she were male. After creating the Mii,
participants were instructed to relax for one minute of baseline physiological
recording. Then the researcher set up the Swordplay game and told them what avatar
they would be using. If the participant was in the customized avatar condition, the
researcher simply selected the aforementioned avatar for her. If the participant was in
the generic avatar condition, as she was choosing the avatar the researcher said, “I
know you made the other Mii, but you’re actually going to be playing with this one.”
It is important to note here that if a participant was assigned to use a generic avatar,
the gender of this avatar matched the gender of the avatar that the participant had
customized.
The researcher then trained the participants on how to play and told the
participants to watch the replays after each round and to keep playing matches until
instructed to stop. He then left the play area and went to a sectioned off area of the
room. After they played for approximately 10 minutes, participants were allowed to
finish the current match and then told to stop playing. The researcher then turned off
the television and gave the participants the iPad for a 15-minute survey that included
the SPQ and measure of presence. After this survey, participants engaged in another
play session and survey, both of which are not included in the present analysis but are
described in Chapter 6.
Measures
Participants responded to all of the questions in the SPQ. The measure of
extended self-presence relating to gender was not included in the analysis because
avatar gender was experimentally manipulated. Items representing each level of the

66
SPQ were combined into composite measures based on the factor analysis. This
process and these measures are described in the Results section.
Participants also responded to a seven-item measure of presence (Fox, et al.,
2009). The items in this measure were combined into a composite measure
(Cronbach's alpha = .87) for analysis. This measure of presence is included in the
analysis as a covariate because the close relationship between presence and self-
presence described earlier implies that the two may be positively related.
Although the generic avatars were pretested to be average-looking, they
mostly appeared to be racially white. In order to control for this confound in the
results regarding extended self-presence, participant race was used as a covariate.
The variable used was a simple binary indicator of whether the participant identified
herself as racially white (41; 64%) or not (23; 36%).
The order of measures in the three surveys was carefully chosen based on the
position of the given survey relative to the play session. In the first survey, before
playing, participants provided only simple demographic information (e.g., age,
education, race). In the second survey, about 20 minutes later, participants responded
to the SPQ and measure of presence. These measures were included in this survey
because it was administered immediately after they played and so the experience was
fresh in their minds. The timing of the additional measures that are not included in the
present chapter, such as the physiological data, was also deliberately unobtrusive.
The physiological data are discussed in Chapters 5 and 6.
Chapter 4: Results
An exploratory principal-axis factor analysis with an oblique rotation (Kim &
Mueller, 1978), was used to determine the SPQ’s factor loadings. This analysis was
confirmatory with respect to the three expected factors, but also exploratory in that the

67
factor loadings determined whether items were retained. The items that loaded poorly
for each given factor or cross loaded between factors were removed, leaving 13 items
that formed three distinct factors. These factors were consistent with the proto, core,
and extended levels of self-presence (Hypothesis 1) and had eigenvalues of 4.71, 1.37,
and 2.41, respectively. Tables 4.1 and 4.2 contain the pattern and structure matrices,
respectively.
Table 4.1. Pattern Matrix of SPQ

Factor

1 2 3
P_arm_stretched
0.84 -0.11 0.01
P_reach_into_game
0.79 -0.04 0.04
P_extension_of_body
0.69 0.03 0.08
P_part_of_Body
0.65 0.19 0.16
P_hand_inside_game
0.62 0.15 0.09
C_arousing 0.11 -0.01 0.74
C_scary
0.13 0.04 0.56
C_sad 0.31 -0.06 0.51
E_skin_color
-0.20 0.72 0.34
E_race
-0.12 0.71 0.22
E_clothing
0.03 0.69 -0.08
E_appearance
0.29 0.68 -0.18
E_name
0.14 0.65 -0.16

Table 4.2. Structure Matrix of SPQ

Factor

1 2 3
P_arm_stretched
0.82 0.08 0.29
P_reach_in_game
0.80 0.15 0.32
P_part_of_Body
0.75 0.36 0.43
P_extend_body
0.73 0.20 0.34
P_hand_in_game
0.69 0.30 0.34
C_arousing
0.37 0.14 0.77
C_scary
0.34 0.17 0.61
C_sad 0.48 0.09 0.61
E_skin_color
0.08 0.74 0.39
E_race
0.12 0.72 0.30
E_appearance
0.38 0.72 0.04
E_clothing
0.15 0.68 0.05
E_name
0.22 0.65 0.00

68
Based on these results, composite measures of the individual levels of self-
presence were created using only the items that were retained in the factor analysis.
Cronbach’s alpha for the proto self-presence measure was .89, for core self-presence
was .77, and for extended self-presence was .80. A zero-order Pearson correlation
analysis, single tailed (because of expected positive correlation) indicated that the
levels of self-presence were positively related (Hypothesis 1), but this relationship
was only nearly significant for core and extended self-presence, as Table 4.3 shows.
Table 4.3. Pearson correlation between composite measures of self-presence.
Presence Proto_Avg Core_Avg Ext_Avg
Presence 1.00
Proto_Avg .60** 1.00
Core_Avg .43** 0.51** 1.00
Ext_Avg .14 0.28* 0.19† 1.00
Note: † p < .10, *p < .05, **p < .01, 1-tailed N = 64

The discriminant validity (Fornell & Larcker, 1981) between presence and the
three levels of self-presence (Hypotheses 2a-c) was established by comparing the
squared correlations between the given pairs of constructs to the average variance
extracted (AVE) for these pairs of constructs, computed with path analysis software
(Ringle, et al., 2005). The AVE values for presence and proto self-presence (0.53 and
0.78, respectively) were higher than the squared correlation between the two
constructs (0.36). The AVE values for presence and core self-presence (0.53 and
1.06, respectively) were higher than the squared correlation between the two
constructs (0.19). The AVE values for presence and extended self-presence (0.41 and
0.91, respectively) were higher than the squared correlation between the two
constructs (0.02). Thus, the three levels of self-presence were each found to be
distinct from presence, supporting Hypotheses 2a-c.

69
Regression analyses were used to test for the effects of avatar customization
and gender consistency on the three levels of self-presence. The avatar customization
independent variable was coded as –1 for generic and 1 for customized, while the
avatar gender was coded as –1 for oppositely gendered and 1 for similarly gendered.
Thus, the interaction term, avatar customization * avatar gender, assigned a value of –
1 to generic, similarly gendered or customized, oppositely gendered avatars and a
value of 1 to customized, oppositely gendered or generic, oppositely gendered avatars.
Because the studies in the previous chapters as well as the present study found the
levels of self-presence to be interrelated, the composite measures of self-presence not
included as the dependent variable were used as covariates. The results for these tests
with respect to extended, proto, and core self-presence can be found in Tables 4.4,
4.5, and 4.6, respectively (in order of hypotheses).
According to Hypotheses 3 and 4, extended self-presence was expected to
relate to avatar customization and gender consistency, respectively. Results of the
regression analysis, Table 4, suggest that Hypothesis 3 was supported but Hypothesis
4 was not. People who used a customized avatar reported higher levels of extended
self-presence. Avatar customization was not a significant predictor of either proto or
core self-presence (Hypotheses 5 and 6, respectively), but gender consistency was for
both of these levels of self-presence (Hypotheses 7 and 8). However, while people
who used a similarly gendered avatar reported higher levels of proto self-presence
than people who used oppositely gendered avatars, the opposite was true for core self-
presence. Namely, people who used a similarly gendered avatar reported lower levels
of core self-presence than people who used oppositely gendered avatars. Thus
Hypotheses 5 and 6 were unsupported, Hypothesis 7 was supported, and Hypothesis 8
was unsupported, but was supported in the opposite direction. Finally, the

70
examination of the interaction between independent variables with respect to the three
levels of self-presence (Research Question 1) was significant only for proto self-
presence. People who used a customized, similarly gendered or generic, oppositely
gendered avatar reported higher amounts of proto self-presence than people who used
a generic, similarly gendered or customized, oppositely gendered avatar. Because of
this interaction effect for proto self-presence, the main effect of avatar gender on
proto self-presence requires further interpretation. To this end, the estimated marginal
means displayed in Table 4.7 and illustrated in Figure 4.2 suggest that the main effect
is indeed genuine.
Table 4.4. Regression Models on Extended (identity) Self-Presence

Beta
t

Cohen's
d size^
Avatar Customization 0.54 5.00 ***
1.34 large
Gender Similarity (female) 0.06 0.53

Interaction (customization * gender) 0.16 1.35

Race (white) 0.21 1.95 †
0.52 medium
Presence 0.14 1.01

Proto (body) Self-Presence 0.12 0.75

Core (emotion) Self-Presence 0.14 1.06

df 56
R
2

0.40

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, >0.2 medium, <.2 small (Cohen, 1988; Rosnow & Rosenthal,
1996)

71
Table 4.5. Regression Models on Proto (body) Self-Presence
Beta t Cohen's d size^
Avatar Customization 0.04 0.36

Gender Similarity (female) 0.24 2.58 *
0.69 medium
Interaction (customization * gender) 0.21 2.11 *
0.56 medium
White 0.04 0.44

Presence 0.52 4.99 ***
1.33 large
Core (emotion) Self-Presence 0.28 2.64 *
0.70 medium
Extended (identity) Self-Presence 0.09 0.75

df 56
R
2

0.55

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, >0.2 medium, <.2 small (Cohen, 1988; Rosnow &
Rosenthal, 1996)

Table 4.6. Regression Models on Core (emotion) Self-Presence
Beta t Cohen's d size^
Avatar Customization -0.20 -1.57

Gender Similarity (female) -0.25 -2.23 *
-0.59 medium
Interaction (customization * gender) 0.11 0.91

White 0.03 0.26

Presence 0.09 0.60

Exttended (identity) Self-Presence 0.14 1.06

Proto (body) Self-Presence 0.39 2.64 *
0.70 medium

df 56
R
2

0.38

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, >0.2 medium, <.2 small (Cohen, 1988; Rosnow &
Rosenthal, 1996)

Table 4.7. Estimated Marginal Means of Proto Self-Presence by Condition

Interaction Condition Mean SD
Customized, same gender
3.41 0.18
Customized, opposite gender
2.63 0.16
Generic, same gender
2.98 0.17
Generic, opposite gender
2.92 0.18

72
Figure 4.2. Estimated Marginal Means of Proto Self-Presence by Condition

Chapter 4: Discussion
The present chapter examines self-presence in a controlled experimental
context using a console-based cartoon-sword-fighting video game. Beyond
supporting the SPQ’s internal consistency and discriminant validity from the concept
of presence, results relating to the manipulated independent variables, avatar
customization and gender consistency, provide additional support for the construct’s
validity and raise some notable questions. Gender consistency but not avatar
customization affected both proto and core self-presence, indicating that the
distinction between customization and personalization warrants investigation.
Further, the interaction effect between the two independent variables with respect to
proto self-presence indicates that being able to use a customized or similarly gendered
avatar does not always contribute positively to the user’s experience. These issues are
examined in greater detail below.
The factor analysis of the SPQ was consistent with the framework of self-
presence, but there are some notable differences between these results and those from
the previous chapters. For one, only three items were retained in the core self-

73
presence factor due to poor factor loadings or cross loadings with other factors. One
reason for this may be that the items retained (arousing, scary/afraid, sad) are more
relevant to the game context than the items dropped (upsetting/angry, surprising,
happy). The sword-fighting activity was certainly arousing, the idea of getting hit by
the opponent is scary, and losing a match could certainly cause sadness. In contrast,
the game is not serious enough to make players angry, and the scenario was static and
repetitive, thus not surprising. One surprise, however, is that happiness did not load
well with the factor because players would presumably feel happy after winning a
match with their avatar. In fact, in a bivariate correlation of this item and the levels of
self-presence, the item is more strongly correlated with the proto self-presence factor
(.50, p < .001) than the core factor (.39, p < .001). This indicates that the experience
of proto self-presence during this type of physically involving game is associated with
happiness during the game. Thus, when responding to the SPQ, participants are
unable to distinguish between their happiness with the game in general and the
relationship between their happiness and events that happen to the avatar in the game.
However, participants may be able to consider this distinction more clearly when
considering emotions that are not as closely tied to the game experience itself, but are
more clearly reflected by interactions between the avatar and the virtual environment.

Regardless of the specific interpretation, this finding illustrates two important
considerations for future use of the SPQ. First, as described in Chapter 2, the SPQ
should be treated as an inventory of questions from which the researcher selects those
that are relevant to the given virtual environment and research questions. The
researcher should also use factor analysis to remove items with cross or weak
loadings. Second, more specific to core self-presence, the wording of these items
could be modified to reflect the specific interactions between the avatar and the

74
virtual environment. For example, instead of “When arousing events happen to your
avatar, to what extent do you feel aroused?” participants could be asked, “When your
avatar struck the opponent, to what extent did you feel aroused?” or “When your
avatar fell off of the platform, to what extent did you feel sad?” This would allow for
a greater level of specificity and control over the measurement. Unfortunately, this
could also lead to a prohibitively large number of questions if the virtual context
includes numerous event types and potential emotional responses. In this case, the
researcher could offer a mix of questions with specific descriptions of events as well
as more general wordings.
The findings supporting the discriminant validity between the three levels of
self-presence and presence represent an important milestone in establishing the
concept of self-presence that the previous chapters were unable to address. Namely,
the concept of self-presence should be treated as distinct from the concept of
presence. While the two are related, future research should not substitute one for the
other but should choose the appropriate construct depending on whether the research
addresses avatar use (self-presence) or the virtual environment itself (presence).
Moving to the predictive relationships, the finding that avatar customization
influenced extended self-presence supports the validity of this level of self-presence.
Through customization, people are able to imbue avatars with characteristics that
reflect their own identities. Thus, compared to generic avatars, customized avatars
are likely to induce a stronger sense that the avatar is a reflection of the self at the
identity level, i.e., extended self-presence. While this finding is unsurprising, it
represents the first confirmation of this level of self-presence in a controlled test.
Unexpectedly, there was no difference in extended self-presence between
people who use similarly gendered and oppositely gendered avatars. This is possibly

75
because the items included in the factor (race, clothing, skin color, name, and
appearance) are not intrinsically related to gender. However, some of these items
may have gendered relationships, such as pink clothing signaling femaleness
4
, and so
alternate explanations should be considered. It is possible that the Miis used in this
study were too cartoonish or childlike and thus the gender differences were not strong
enough of a manipulation to affect extended self-presence. Future research may find
that gender consistency induces extended self-presence in virtual environments where
avatar gender is more obvious or sexualized, such as Second Life.
The finding that proto self-presence was positively influenced by avatar
gender consistency but not customization reflects the importance of the distinction
between personalization and customization, as introduced in Chapter 2. Given that
gender is one of the most fundamental aspects of human identity (Bem, 1981), even
when able to customize the attributes of an oppositely gendered avatar, participants
may not have felt as though they were able to personalize it. In contrast, using a
similarly gendered avatar, regardless of customization, may facilitate more of a
personal connection than using an oppositely gendered avatar. Thus, the difference
between using a similarly gendered versus oppositely gendered avatar parallels the
difference between avatar types in Study 3 of Chapter 2: avatar with the user’s image
on the face versus avatar customized with in-world tools only. Namely, the former
category of avatar facilitates more of a personal connection. And just as in Study 3 of
Chapter 2, participants who had this more personal connection in the present study, by
using a similarly gendered avatar, reported higher levels of proto self-presence.

4
Hairstyle is another gendered identity characteristic that is important in avatar customization
(Ducheneaut, et al., 2009) and thus should be included in future iterations of the SPQ.

76
The cause of this relationship between avatar personalization and proto self-
presence may relate to the physical presentation of the avatars’ physical attributes. In
Study 3 of Chapter 2, the personalized avatar looked more like the user, because it
included a picture of the user’s face, than the avatar customized with in-world tools.
Similarly, in the present study, because avatar gender is much more obvious than the
other customizable features (e.g., skin color, name, clothing), regardless of
customization condition, the participants in the similarly gendered avatar condition
probably felt more physically similar to their avatars than participants in the
oppositely gendered condition. This indicates that personalization is reflected by
physical similarity to the avatar and that such similarity facilitates a body-level
connection to the avatar. This is consistent with neuroscientific research that shows
that the perceptual integration of non-body parts into body schema is stronger when
those non-body parts resemble the actual body part in appearance or are oriented
realistically (Armel & Ramachandran, 2003; Ehrsson, et al., 2004; Tsakiris &
Haggard, 2005). Future self-presence research should consider this relationship
between proto self-presence and personalization or physical similarity of user to
avatar.
The finding that core self-presence was negatively influenced by avatar gender
consistency was not expected, but can be interpreted by also considering the effect of
an avatar’s physical attributes as well as basic gender stereotypes. Specifically,
because all of the participants were female, those in the opposite gender condition
used male avatars. Although these participants felt a less personal connection to their
avatars, as illustrated by the previously described finding, these participants may have
felt as though they were taking a male role in the game. Females who use male
characters in video games have been found to engage in more stereotypically

77
masculine behaviors within the game, such as combat behaviors that indicate
competitiveness (Huh & Williams, 2010), and so perhaps these female participants
who used male avatars were induced to feel more competitive and thus more
emotionally affected by events in the game.
Lastly, the interaction effect between the avatar assignment conditions with
respect to proto self-presence may reflect a nuance in this particular study’s design.
All participants designed an avatar within the appropriate gender condition to ensure
similarity in experience across conditions, so those who were later assigned to use a
generic avatar were aware that they were not using the avatar they had created. This
likely was more salient for the participants who were in the consistent gender
condition because they would have established a stronger connection with the avatar
they built than participants in the opposite gender condition, for whom building the
avatar was less an expression of identity due to the gender restriction. Thus, by
assigning these participants to use an avatar they had not built, those who customized
a similarly gendered avatar would have been more disappointed than those who
customized an oppositely gendered avatar. Further, those participants who customized
an oppositely gendered avatar may not have felt comfortable enough to express or
consider any of their personal identity characteristics that are oppositely gendered.
Although virtual worlds and video games present environments in which many people
experiment with gender switching (MacCallum-Stewart, 2008; Roberts & Parks,
1999; Yee, 2008), this activity usually takes place behind a veil of anonymity. In the
present study, the participants were separated from the researcher by a narrow room
divider and their behaviors were obviously being examined. Thus, participants who
used customized, oppositely gendered avatars probably felt more uncomfortable than
those who used generic, oppositely gendered avatars. These points together indicate

78
that using a generic, similarly gendered avatar or a customized, oppositely gendered
avatar detracts from the player’s connection to the avatar. As the results reflect, this
reduced connection to the avatar hinders proto self-presence. This interaction effect
may relate to interaction effects in subsequent chapters of this dissertation that report
results from the same study.
Overall, this chapter provides the first experimental examination of the
internal consistency and validity of self-presence and the SPQ. Given that only
female participants were included in the study and that males and females tend to play
video games differently (Huh & Williams, 2010; Williams, et al., 2008), the results
presented here may not be generalizable to males. Still, results from the examinations
presented in the previous chapters, which included both males and females, are
consistent with the present results, especially relating to the internal consistency of the
SPQ. Thus, this chapter concludes the internal consistency testing that will be
presented in this dissertation. There appears to be enough support for the claim that
the SPQ is a reliable measure of the conceptual framework of self-presence that the
SPQ can be used without replication of such tests. However, as indicated throughout
the previous chapters, researchers should still tailor the SPQ to the specific virtual
environment in question and then confirm appropriate factor loadings for the items.
The validity tests presented in the chapter are strong, due to the experimental
manipulation of avatar assignment conditions, but this does not conclude the validity
testing in this dissertation. In the subsequent chapters, physiological data recorded
during this experiment will be compared to the SPQ. These tests will serve both as an
indication of the SPQ’s validity as well as a reflection of the types of research
questions that can be examined using the concept of self-presence.

79
Chapter 5: Physiological Correlates of Self-Presence During Avatar Use

Chapter 5: Abstract
The present chapter uses a psychophysiological approach to examine self-
presence and the Self-Presence Questionnaire (SPQ) during avatar use. This
examination occurs within the same experimental context described in the previous
chapter. Experimental condition and responses to the SPQ are compared to two
physiological indicators of arousal, heart rate and electrodermal activity, as well as
two physiological indicators of emotional valence, zygomatic (smiling) and
corrugator (frowning) electromyogram activity. Use of a generic, oppositely
gendered avatar was found to increase negative emotional responses. Use of
customized, similarly gendered or generic, oppositely gendered avatars, compared to
generic, similarly gendered or customized, oppositely gendered avatars, was found to
reduce arousal (i.e., electrodermal activity). And both proto and core self-presence
were positively related to positive emotional valence. While these results provide an
interesting contribution to the understanding of psychophysiology during avatar use,
the physiological responses do not necessarily indicate direct or causal relationships
with self-presence because of unavoidable limitations in the present design. This
insight is useful for identifying an appropriate design for testing causal relationships
between self-presence and a wider range of physiological responses, as presented in
the following chapter.

Self-presence is a psychological experience that depends on the ways and
extent to which people connect to their virtual self-representations. However, self-
presence is also manifested in the entire body and thus is also a physiological

80
experience. Just as Damasio (1994, 1999) describes the self as a highly interconnected
system comprised of the brain and body, the experience of self-presence is reflected in
the body’s reactions to using a virtual self-representation. Specifically, an
individual’s physiological state during avatar use may vary in ways that are
predictably consistent with variations in this individual’s experience of self-presence.
This chapter examines such relationships between self-presence and physiology,
providing an additional test of the construct of self-presence and the Self-Presence
Questionnaire (SPQ) measurement tool.
The field of psychophysiology contributes to an understanding of
psychological processes through the analysis of physiological states that reflect such
processes (Hugdahl, 1996). For example, early research on skin conductance,
conducted in the late 19
th
century, examined relationships with simple stimuli, such as
the way this physiological signal indicates the feeling of surprise (Dawson, Schell, &
Filion, 2000). More recent research using this physiological indicator ranges from the
examination of differences during affective conditioning (Dawson, Rissling, Schell, &
Wilcox, 2007) to symptoms in schizophrenic patients (Dawson & Schell, 2002), to a
wide variety of other applications (Dawson, et al., 2000).
In the past quarter century, psychophysiology has become a viable approach to
the study of communication and media (Lang, Potter, & Bolls, 2009; Ravaja, 2004).
Research on traditional media have used such methods to examine a variety of effects,
such as arousal differences between high and low sensation seekers when watching
public service announcements (Lang, Chung, Lee, Schwartz, & Shin, 2005),
differences in cardiac orienting responses when controlling or not controlling the
onset of emotional images (Wise & Reeves, 2007), arousal differences when listening
to audio messages of varied complexity (Potter & Choi, 2006), and emotional valence

81
and arousal differences based on an interaction of participant mood and message
presentation via text versus video modality (Ravaja, Saari, Kallinen, & Laarni, 2006).
There are also numerous studies on psychophysiological effects of using interactive
media, such as arousal differences related to video game violence (Ivory &
Kalyanaraman, 2007) and immersion (Persky & Blascovich, 2007), emotional valence
differences when accomplishing or not accomplishing goals in a video game (Ravaja,
Saari, Salminen, Laarni, & Kallinen, 2006), and cortical activity in specific brain
structures when playing violent video games (Weber, Ritterfeld, & Mathiak, 2006).
Out of the numerous psychophysiological methods, many of the
communication and media studies in this area focus on arousal and emotional valence.
This is likely because the measurement and interpretation of these physiological
processes is relatively straightforward compared to the many others that are more
common in the academic field of Psychophysiology (Ravaja, 2004). Physiological
arousal, an outcome of the sympathetic nervous system’s “fight or flight” response
(Hugdahl, 1996), is commonly associated with the emotions of excitement, anger, or
fear. Emotional valence describes the positivity or negativity of an emotion. For
example, joy and excitement are positive emotions while disgust and sadness are
negative. While arousal and emotional valence independently provide information
about an individual’s state, certain combinations of the two provide other unique
information. For example, decreases in some measures of arousal combined with
negative emotional valence is associated with increased attention, which is of
particular interest to communication and media researchers (Ravaja, 2004). The
present dissertation utilizes this understanding of arousal and emotional valence to
test the validity of self-presence and the SPQ.

82
Self-presence and psychophysiology
The avatar-assignment variables presented in the previous chapter’s
experiment – avatar customization, gender consistency, and the interaction between
these variables – may affect physiological responses during avatar use. These
variables were found to influence self-presence. Specifically, avatar customization
increased extended (identity-level) self-presence, gender consistency increased proto
(body-level) and decreased core (emotional-level) self-presence, and the interaction
between customization and consistency increased proto (body-level) self-presence.
When the experience of self-presence is strong, interactions between the avatar and
the virtual environment are treated to some extent as unmediated, and this may
increase the strength of physiological responses during avatar use. Thus, people who
are induced to feel a stronger connection to their avatar, via avatar customization,
gender consistency, or the interaction between these variables, may exhibit more
physiological arousal or stronger emotional responses during avatar use. However,
these avatar-related manipulations may affect facets of the avatar-use experience,
besides self-presence, that influence physiological responses in other ways. For
example, because assignment to a generic avatar may be disappointing (see Chapter
4), players may be more agitated and thus exhibit more arousal or negatively valenced
emotional responses during avatar use. Similarly, assignment to an oppositely
gendered avatar may inhibit enjoyment and thus increase negatively valenced
emotional responses. Given this conflicting set of expectations, the relationships
between these avatar assignment variables and physiological responses during avatar
use are explored through the following set of research questions:
Research Question 1: Does assignment to a customized or generic avatar influence the
strength of physiological arousal and emotional responses during avatar use?

83
Research Question 2: Does assignment to a similarly gendered or oppositely gendered
avatar influence the strength of physiological arousal and emotional responses during
avatar use?
Research Question 3: Does the interaction between avatar customization and gender
consistency influence the strength of physiological arousal and emotional responses
during avatar use?
Apart from the influence of avatar-related variables, the experience of the
three levels of self-presence may be distinctly related to physiological responses
during avatar use, but the temporal order of this relationship is unclear. Proto self-
presence occurs only during avatar use, while core and extended self-presence can
persist afterwards (Chapter 1). Physiological responses during avatar use are
measured in real time, but the measurement of self-presence must occur after initial
avatar use. Therefore, the psychological experience of self-presence may affect
physiological states during avatar use, or vice versa. Regardless of causality, an
examination of the relationship between these two constructs provides insights into
self-presence and the SPQ.
When someone experiences proto self-presence, she feels as if she were using
some part of her own body in the virtual environment and thus experiences more
arousal and stronger emotional responses than if she did not feel such a connection to
an avatar. For example, consider an individual using his own hand or a pair of tongs
to pick up and move a lit firecracker. When using his hand, he will certainly be more
aroused and disturbed than when using the tongs. This difference obviously results
from his concern about injuring his hand. Similarly, when an avatar is integrated into
body schema, the use of that avatar may lead to more arousal and stronger emotional
responses. The following hypotheses reflect this reasoning.

84
Hypothesis 1: Proto self-presence will be positively related to physiological arousal
during avatar use.
Hypothesis 2: Proto self-presence will be positively related to the strength of
emotional responses during avatar use.
Core self-presence, by definition, reflects the extent of emotional responses
during specific interactions between the virtual self-representation and objects in the
virtual environment. Some of the items in the measure of core self-presence relate
more to arousal, while others relate more to emotional valence. The items relating to
surprise, arousal, and fear fall into the former category, while the items relating to
happiness, sadness, disgust, and anger fall into the latter. As described in the previous
chapter, not all of these items are likely to be relevant in a given virtual context. The
specific spread of items that are included in this measure may influence the extent to
which core self-presence is related more to physiological arousal or emotionally
valenced responses. Still, both types of responses can result from the same emotional
experiences, and so regardless of the specific items, core-self presence may be
positively related to both types of responses.
Hypothesis 3: Core self-presence will be positively related to physiological arousal
during avatar use.
Hypothesis 4: Core self-presence will be positively related to the strength of
emotional responses during avatar use.
The experience of extended self-presence differs from the other two levels of
self-presence in that extended self-presence is not likely to be related to arousal
because identity-level connections are experienced as memories of personal attributes,
not visceral feelings, and thus are usually unrelated to the sympathetic nervous
system’s “fight or flight” responses (Damasio, 1999). However, extended self-

85
presence may be related to emotionally valenced physiological responses because
people feel strong attachments to their identities, perhaps because identity and
reputation were important contributors to the survival of early social humans (Evans
& Zarate, 1999). Thus, when someone feels a strong identity-level connection to a
virtual self-representation, this person may experience positive emotions when the
avatar is in positive situations and negative emotions when the avatar is in negative
situations.
Hypothesis 5: Extended self-presence will be positively related to the strength of
emotional responses during avatar use.
Chapter 5: Method
The 2x2 experiment described in the previous chapter is the same one used
here. The Method section in the previous chapter articulates the design, participants,
game-related materials, and general procedure, and therefore will not be presented
here in detail. Instead, the present description will focus on the physiological
measurements and procedures that were included in this experiment.
Measures
The two measures of arousal used in the study were heart rate (HR) and
electrodermal activity (EDA), though HR can also determine emotional valence
(Lang, 1994). HR is essentially the frequency at which the heart expands and
contracts, pumping blood throughout the body’s cardiovascular system, and is thus
measured in beats per minute (Hugdahl, 1996). The relationship between HR and
arousal is intuitive because when most people engage in vigorous activity, such as
exercise, the increase in HR is obvious. Further, without engaging in any activity, HR
increases in stressful situations, such as giving a public speech or engaging in a motor
task that will yield rewards if completed correctly and quickly (Hugdahl, 1996). From

86
an evolutionary perspective, the relationship between HR increases and physical
activity or stress is simply the sympathetic nervous system’s “fight or flight” response
to situations in which the body’s muscles require increased oxygen and thus blood
flow, as delivered by the cardiovascular system.
EDA, defined as, “the ability of the skin to conduct electrical current”
(Hugdahl, 1996, p. 101), is one of the most widely used psychophysiological
measures (Dawson, et al., 2000). When sweat glands on the palms of the hands and
bottoms of the feet become active, the amount of current passed through two
electrodes placed on the skin’s surface increases (Hugdahl, 1996, p. 101). Thus, EDA
is measured in units of electrical conductivity, microsiemens (Dawson, et al., 2000).
While changes in EDA may not be as obvious as changes in HR, people may be
aware of the experience of sweaty palms when they are nervous. However, measures
of EDA are sensitive to small changes in moistness that are often not perceivable.
These changes accompany changes in stress or nervousness at a 1-3 second delay,
which is one reason why measures of EDA are commonly incorporated into
polygraph tests for lie detection (Matte, 1996). The evolutionary reasoning for the
relationship between EDA increases and stress or nervousness is that the increased
sweat-gland activity on the feet and hands increases grip strength, balance, mobility,
and protection against wounds, which are all important during a “fight or flight”
response (Hugdahl, 1996).
Emotional valence, the extent to which someone experiences positive or
negative emotions, is reflected in facial muscle activity. While there are numerous
combinations of facial muscle activity (i.e., facial expressions) that indicate nuanced
differences in an individual’s emotional experience (Ekman, 1971, 1993), there are
also simple relationships between some facial muscles and basic emotional valence.

87
Specifically, contraction of corrugator (eyebrow) muscles indicates negative emotion
while contraction of zygomatic (cheek) muscles indicates positive emotion (Hugdahl,
1996). One common method of measuring such muscle activity is the
electromyogram (EMG), in which electrodes are placed on the skin to record “action
potentials generated when the fibers of a muscle contracts” (Hugdahl, 1996, p. 335),
measured in millivolts. These electrical signals generally increase as the strength of
the muscle contraction increases, and thus provide a sensitive measure of muscular
responses to stimuli. In the case of facial EMG, the signal from the corrugator and
zygomatic muscles indicate the extent to which an individual experiences negative or
positive emotions, respectively.
As described in the previous chapter, composite measures of self-presence
were constructed from responses to the SPQ. The present analyses utilize the same
measures included in that chapter.
Materials

All physiological data were acquired using the 4-channel MP36 system and
the Biopac Student Lab software from Biopac Systems, Inc. (www.biopac.com). An
electrocardiogram (ECG), the signal from which HR is calculated, was recorded from
cloth electrodes placed on the center of the left forearm and the medial side of the left
ankle just behind the fibula, and with a ground electrode placed on the left wrist.
Each of these locations was prepped with an alcohol swab shortly before application.
The EDA electrodes were placed on the distal phalanges of the index and middle
fingers of the non-dominant (left) hand. The zygomatic and corrugator EMG
electrodes were placed on the left cheek and forehead, with the ground electrodes on
the temple and behind the ear, respectively. Each of these locations was prepped with
Lemon Prep abrasive skin prepping lotion (by Mavidon Corp) as well as an alcohol

88
swab. The zygomatic EMG recording used 8mm silver-silver chloride electrodes
filled with a 0.05 molar isotonic NaCl paste because they were smaller and less likely
to fall off the participants’ faces than the cloth electrodes. An athletic headband was
placed around the forehead to stabilize the corrugator electrodes. Images of the
electrode placement on the arm and face can be found in Figure 5.1.
Figure 5.1. Electrode placement on the arm and face.

After recording, the raw physiological data was transformed using the
AcqKnowledge 4.1 software, also from Biopac Systems, Inc. The ECG output
was converted into HR using the software’s rate finder function. The EMG data
were transformed using the integrate function to average over 100 samples using
the root mean square method with no baseline removal. There were no
transformations conducted on the EDA data. Raw and transformed data for an
example participant can be found in Figures 5.2 and 5.3, respectively.

89
Figure 5.2. Example of raw physiological output

Figure 5.3. Example of transformed physiological output (EDA not transformed)

Procedure
Only the procedures that relate to the physiological data recorded during
avatar use are described here in detail. After the participants read the IRB-approved
information sheet and completed the initial survey, the researcher informed the
participants that he was going to attach some sensors to their arms and face. He

90
started with the ECG sensors, first telling the participants that he was going to use an
alcohol swab and then wiping the relevant areas of the participants’ left arm. He then
gave the participants an alcohol swab and instructed them to wipe the back of their
left ankle. He pointed to the area just behind his own fibula so that the participants
knew where to wipe. He then placed the adhesive cloth electrodes on the areas that he
had wiped on the arm and gave the participants an electrode to place on their ankle.
Next, he clipped the lead of the wire from the MP36 to the electrodes on the arm and
then instructed the participants to do the same on the ankle. Finally, he placed a piece
of medical tape over the electrodes and instructed the participants to do the same on
the ankle. The researcher then placed the EDA electrodes on the participants’ left-
hand index and middle finger tips. He secured these electrodes with medical tape and
clipped the leads to them. Next, he told the participants that he was going to exfoliate
a small area on their cheek and forehead. He used a Q-tip to lightly rub Lemon Prep
on the areas on which the facial EMG electrodes would be placed and then removed
the residue with a cotton swab. Next, after telling the participants to expect the smell
of alcohol, he used an alcohol swab to wipe the participants’ left cheek and forehead
where he had exfoliated. He then placed the appropriate electrodes on these locations.
Finally, he placed the zygomatic ground electrode behind the left ear and used a
headband to secure the corrugator electrodes.
The researcher then commenced with the study procedure as described in the
previous chapter. The participants spent approximately 5 minutes designing either a
similarly gendered or oppositely gendered avatar, depending on condition. Then
participants were instructed to relax, take a deep breath, and close their eyes for one
minute of baseline physiological recording.

91
The participants were then instructed to play repeated matches of the
Swordplay game. Each match included multiple rounds. In order to win a match,
either the participant or the computer opponent had to win two rounds. Most matches
were comprised of two rounds because participants won 90% (s.d. = 9.54) of all
rounds. If the participant lost one round and won one round in a match, the match
went to a third round. In the case where the participant reached the round time limit
(60 seconds) without winning or losing, the round was marked as a tie. In the rare
case where the participant won, lost, and tied one round, the match went to a fourth
“sudden death” round in which the play arena size shrunk and only one hit was
required to win. The researcher marked the start and finish of each round in the
physiological recording software. The mean round time was 21 seconds (sd = 4.8).
Because participants played repeated matches for a set amount of time (approximately
10 minutes), the total number of rounds varied between participants (m = 14.70, sd =
1.57).
After 10 minutes of play, the researcher waited until the participants had
finished the current match, then instructed the participants to stop playing, and turned
turned off the television. Participants then completed a 15-minute survey that
included the SPQ, engaged in another session that will be examined in the following
chapter, and completed a 5-minute survey that included questions about technology
habits. The researcher then removed the physiological recording equipment and
debriefed the participants on the purpose of the study.
Metrics
Based on procedures specific to this experiment as well as the nature of the
physiological measures, a number of metrics were constructed to test the present
hypotheses. Given the variance in the time per round, number of rounds per match,

92
and total number of rounds, the physiological data for each participant was averaged
across all rounds. This constitutes a tonic physiological metric, or an average over
time (Hugdahl, 1996). The one-minute baseline measurement was also averaged into
tonic metrics.
Because individual physiological differences can be very large (Hugdahl,
1996) and the relationship between psychological and physiological events not very
high (Ravaja, 2004), metrics were also constructed to maximize the relationship
between the various stimuli and variance in each physiological indicator. One of the
most basic ways to control for individual differences in physiological responses is to
take a baseline measure of participants at rest and then compare this measure to
physiological responses to stimuli (Kaiser, 1989). In other words, this method
controls the extent to which individuals’ initial physiological state affects their
physiological responses during the experiment. There are two ways in which this
metric can be constructed, as an absolute change score (value – baseline) or a
percentage change score ((value – baseline) / baseline). The choice of appropriate
metric depends on whether the metrics are dependent on the baseline values, which
can be determined by evaluating a scatterplot of the metrics and baseline values
(Kaiser, 1989). The superior metric is the one that is less dependent on baseline.
Using this method, absolute change was found to be the appropriate metric for both
zygomatic EMG and corrugator EMG.
Another common method of controlling for individual differences in tonic
measures, which has been found to produce more accurate results than the comparison
of raw values, is to consider the participants’ range of physiological responses
(Lykken, Rose, Luther, & Maley, 1966). This method is particularly useful for
measures in which there is significant variance in the minimums and maximums of

93
which participants are capable, such as HR and EDA. One metric for such a
correction is calculated by dividing the difference between the physiological value in
question and participant’s minimum value by the participant’s range of values for the
given measure (Lykken, et al., 1966). This metric is expressed in the following
formula: (value – minimum) / (maximum – minimum). In the present analysis, the
“value” is the average of the tonic measures taken for the given segment, the
“minimum” is the lowest value recorded from the start of the baseline recording to the
completion of the experiment, and the “maximum” is the highest value from this same
segment.
Another method of controlling for individual differences in tonic measures,
which provides improvements over range correction in some cases, is to standardize
physiological output into z scores (Ben-Shakhar, 1985). Such a metric is calculated
by dividing the difference between the individual score and the overall mean for a
given physiological measure by the overall standard deviation of that measure, as
follows: (value – mean) / standard deviation. Like range correction, this metric has
been used with measures of HR and EDA (Markovsky, 1988). Using the same
approach as Ben-Shakhar (1985) to compare range-correction and z-score
normalization, both types of metrics were created for HR and EDA and then the
strength of results dictated which metric was retained. This analysis indicated that z-
score normalization yielded stronger results for HR, but range correction yielded
stronger results for EDA.
As an additional means of controlling for individual differences, a number of
covariates were included in the present analysis. The frequency of playing video
games, measured on a 5-point scale in response to the question “How often do you
play video games”, was included because people who are more familiar with a type of

94
stimuli are likely to habituate to such stimuli more quickly than people for whom the
stimuli is novel (Gati & Ben-Shakhar, 1990). The percentage of rounds that the
participants won while playing the game was included because physiological
responses have been found to differ depending on whether the player wins or loses
(Ravaja, Saari, Salminen, et al., 2006). As in the previous chapter, a binary indicator
of whether the participant identified herself as racially white or not was included as a
covariate because of the confound between race and using a generic avatar. This
covariate was also useful because it reduced variance in physiological responses due
to race (Berardesca & Maibach, 1996; Johnson & Landon, 1965; Liao, Barnes,
Chambless, & Simpson, 1995; Vrana & Rollock, 2002). The measure of presence
was also included as a covariate because of its relationship to the concept of self-
presence, as described in Chapter 1.
Chapter 5: Results
The metrics described above were used in regression analyses to examine the
research questions and hypotheses. As in the previous chapter, the avatar
customization independent variable was coded as –1 for generic and 1 for customized,
while the avatar gender was coded as –1 for oppositely gendered and 1 for similarly
gendered. Thus, the interaction term, avatar customization * avatar gender, assigned a
value of –1 to generic, similarly gendered or customized, oppositely gendered avatars
and a value of 1 to customized, oppositely gendered or generic, oppositely gendered
avatars. These three independent variables – avatar customization, gender
consistency, and the interaction between the two – as well as the covariates described
above were entered into the regression analyses as predictors. In four separate
regressions, each of the physiological indicators – HR, EDA, zygomatic EMG, and
corrugator EMG – were entered as the dependent variables. This first series of

95
regression models (Model 1 in Tables 5.1-5.4) provide tests for Research Questions 1-
3, which focus on the effects of the manipulated avatar-assignment variables. In order
to test Hypotheses 1-5, which focus on the physiological correlates of self-presence,
the composite measures of the three levels of self-presence were entered as predictors
into a second regression model (Model 2 in Tables 5.1-5.4) via block entry.
Table 5.1. Regression Models on Heart Rate

Model
1

Cohen's
Model 2

Cohen's
Beta t d size^ Beta t d size^
Avatar customization
0.21 1.48

0.29 1.69
†
0.48 med
Gender consistency
-0.03 -0.20

0.00 -0.01

Interaction
(customization * gender)
0.03 0.25

0.15 1.00

Race (white)
-0.05 -0.35

0.02 0.16

Presence 0.20 1.43

0.40 2.14
*
0.61 med
Video game frequency -0.06 -0.44

-0.01 -0.08

Percent win 0.13 0.94

0.15 1.06

Proto self-presence

-0.18 -0.91

Core self-presence

-0.12 -0.69

Extended self-presence

-0.14 -0.81

df 53 50
R
2

0.08

0.14

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2 small

96
Table 5.2. Regression Model on Electrodermal Activity

Model 1 Cohen's
Model 2 Cohen's
Beta t d size^ Beta t d size^
Avatar customization
0.02 0.14

0.06 0.40

Gender consistency
0.08 0.62

0.13 1.00

Interaction
(customization *
gender) -0.30 -2.46 * -0.66 med -0.31 -2.26 * -0.63 med
Race (white)
-0.20 -1.64

-0.20 -1.52

Presence -0.05 -0.37

-0.01 -0.08

Video game frequency 0.26 2.05
*
0.55 med 0.23 1.74
†
0.48 med
Percent win 0.18 1.45

0.17 1.36

Proto self-presence

-0.15 -0.81

Core self-presence

0.21 1.30

Extended self-presence

-0.02 -0.15

df 55 52
R
2

0.22

0.25

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2 small

Table 5.3. Regression Model on Zygomatic EMG

Model 1 Cohen's
Model 2 Cohen's
Beta t d size^ Beta t d size^
Avatar customization
0.07 0.53

0.11 0.66

Gender consistency
-0.13 -0.99

-0.16 -1.20

Interaction
(customization *
gender) -0.07 -0.54

-0.24 -1.67 p=.10 -0.46 med
Race (white)
-0.04 -0.28

-0.09 -0.68

Presence -0.05 -0.33

-0.35 -1.98
†
-0.55 med
Video game frequency -0.16 -1.19

-0.24 -1.83
†
-0.51 med
Percent win -0.04 -0.32

-0.04 -0.32

Proto self-presence

0.36 1.89
†
0.53 med
Core self-presence

0.27 1.62
p=.11
0.45 med
Extended self-presence

-0.07 -0.41

df 55 52
R
2

0.06
0.21

Note: † p < .10, *p < .05, **p < .01, ***p < .001

97
Table 5.4. Regression Model on Corrugator EMG

Model 1 Cohen's
Model 2 Cohen's
Beta t d size^ Beta t d size^
Avatar customization
-0.20 -1.56

-0.17 -1.11

Gender consistency
-0.25 -2.05
*
-0.56 med -0.19 -1.38

Interaction (custom *
gender)
0.25 2.02 * 0.55 med 0.27 1.94 † 0.54 med
Race (white)
0.14 1.10

0.14 1.05

Presence -0.20 -1.58

-0.11 -0.66

Video game frequency -0.08 -0.66

-0.10 -0.76

Percent win -0.20 -1.59

-0.21 -1.67

Proto self-presence

-0.25 -1.34

Core self-presence

0.18 1.11

Extended self-presence

0.03 0.18

df 54 51
R
2

0.21
0.25

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2 small

Results from the first series of regression models indicate that while avatar
customization did not influence any of the physiological measures, gender
consistency and the interaction between avatar customization and gender consistency
did. Specifically, participants who used similarly gendered avatars exhibited less
corrugator EMG activity (negative emotional responses) than participants who used
oppositely gendered avatars. The interaction between avatar customization and
gender consistency affected both EDA and corrugator EMG. Participants who used
customized, similarly gendered or generic, oppositely gendered avatars exhibited less
EDA (arousal) and higher corrugator EMG activity (negative emotional responses)
than participants who used generic, similarly gendered or customized, oppositely
gendered avatars. While the mean values of EDA (Table 5.5) show a balanced
interaction effect for the former finding, the mean values of corrugator EMG (Table

98
5.6) indicate that this latter effect was driven mostly by high responses from
participants using generic, oppositely gendered avatars. Further, these means also
illustrate virtually no difference caused by gender consistency within participants who
used customized avatars, indicating that the main effect of avatar gender-consistency
on corrugator EMG also results from large responses in participants using generic,
oppositely gendered avatars. Further, the higher standard deviation for this condition
indicates that a small portion of the participants in this condition drove this effect, as
illustrated in Figure 5.4. Overall, these results provide some insight into Research
Questions 1-3, though neither the HR measure of arousal nor the zygomatic EMG
measure of positive emotional responses were affected by any of the manipulated
avatar-assignment variables.
Table 5.5. EDA descriptives by condition (range corrected scores)
Interaction Condition Mean SD
Customized, same gender
0.42 0.10
Customized, opposite gender
0.47 0.11
Generic, same gender
0.49 0.13
Generic, opposite gender
0.42 0.11

Table 5.6. Corrugator EMG descriptives by condition (millivolts * 100)

Interaction Condition Mean SD
Customized, same gender
0.50 0.42
Customized, opposite gender
0.48 0.46
Generic, same gender
0.35 0.40
Generic, opposite gender
0.96 1.26

99
Figure 5.4. Corrugator EMG scatterplot

Results from the second series of regression models indicate minimal support
for Hypotheses 1-5. Proto self-presence was not related to any of the physiological
measures at the p < .05 level. However, the relationship between proto self-presence
and zygomatic EMG (positive emotional valence) approached significance (p = .06).
Thus, Hypothesis 1 was unsupported and Hypothesis 2 was nearly supported for
positive emotional valence. Core self-presence was not related to any of the
physiological measures at the p < .05 level. However, the relationship between core
self-presence and zygomatic EMG approached significance (p = .11). Thus,
Hypothesis 3 was unsupported and Hypothesis 4 was nearly supported for positive
emotional valence. The relationships between zygomatic activity and both proto and
core self-presence are in the positive direction (Table 5.3). Finally, extended self-
presence was not related or nearly related to any of the physiological measures and
thus Hypothesis 5 was unsupported.
Chapter 5: Discussion
The present chapter provides a psychophysiological examination of avatar use.
Results from a 2x2 experiment indicate that use of a generic, oppositely gendered

100
avatar increases negative emotional responses. Further, use of customized, similarly
gendered or generic, oppositely gendered avatars, compared to generic, similarly
gendered or customized, oppositely gendered avatars, reduces arousal (i.e.,
electrodermal activity) during avatar use. Controlling for these causal effects, both
proto and core self-presence tended to be positively related to positive emotional
valence (i.e., smiling) during avatar use. Together, these results provide insights into
the relationships between avatar characteristics, self-presence, and physiological
responses.
The increase of negative emotional valence caused by using a generic,
oppositely gendered avatar is not the direct result of a feeling of connection to the
avatar, but instead the enjoyment of playing the game with this type of avatar. As
described in the previous chapter, participants in the generic condition most likely felt
disappointed because they were unable to use the avatar that they had created.
Further, participants who used an oppositely gendered avatar may have felt
uncomfortable or experienced cognitive dissonance (Festinger, 1957) because, when
given a choice, most people prefer to select similarly gendered avatars (Huh &
Williams, 2010; Hussain & Griffiths, 2008). Together, these two hindrances to
enjoyment most likely caused some of the participants to feel more negatively about
the avatar use experience and thus frown more. However, this conclusion should be
treated carefully because the variance in negative emotional responses for participants
in this condition was relatively high compared to the other conditions, and so a small
portion of the participants in this condition drove this result.
The interaction effect on arousal was more reliable. As described in the
previous chapter, participants who used customized, similarly gendered or generic,
oppositely gendered avatars, compared to generic, similarly gendered or customized,

101
oppositely gendered avatars, reported higher levels of proto self-presence. These
same groups exhibited less arousal (i.e., electrodermal activity) during avatar use, and
although this effect was not mediated by proto self-presence (i.e., it remained
significant in Model 2 of Table 5.2), the explanation for both interaction effects may
be similar. Namely, participants who customized similarly gendered avatars probably
felt more connected to those avatars than those who used customized, oppositely
gendered avatars because the act of customizing was not private, as it often is during
most avatar use. Further, as described above, assignment to a generic, similarly
gendered avatar was likely more disappointing than assignment to a generic,
oppositely gendered avatar. Thus, participants who used generic, similarly gendered
or customized, oppositely gendered avatars would have felt more uncomfortable
during avatar use. EDA responses are higher when people undergo aversive stimuli-
induced stress (Courtney, Dawson, Schell, Iyer, & Parsons, 2010; Lovibond, 1992),
and thus it follows that these uncomfortable participants would have exhibited more
arousal and even perhaps more vigorous movement during the game. This implies
that this interaction effect is not directly related to the connection to the avatar, but
instead to the general experience of playing the game with the given type of avatar.
The findings that both proto and core self-presence were positively associated
with positive emotional valence (zygomatic EMG) is consistent with the conceptual
framework of self-presence, but may result from a direct or indirect relationship
between the connection to the avatar and the game experience. Further, this
relationship may be causal in either direction between self-presence physiological
responses. If the relationship was direct and causal from self-presence to physiology,
then participants who experienced more proto and core self-presence responded to
events in the game as if the avatar was more part of the self. In other words, good

102
things that happened to the avatar also happened to the self, and so these participants
smiled more. If the relationship was direct and causal from physiology to self-
presence, then participants who smiled more felt more connected to their avatars’
successes and thus reported higher levels of proto and core self-presence. If the
relationship was indirect and causal from self-presence to physiology, then the
participants who experienced higher levels of proto and core self-presence simply had
more fun playing the game and thus smiled more. If the relationship was indirect and
causal from physiology to self-presence, then the participants who smiled more had
more fun for a reason besides self-presence, but were likely to report higher levels of
proto and core self-presence for some other reason. All of these explanations are
valid because the physiological outputs and experience of self-presence occurred
simultaneously, though the former was measured before the latter.
The lack of a relationship between any level of self-presence and corrugator
EMG illustrates another notable limitation of the present analysis. Namely, self-
presence should be related to positive as well as negative emotional responses during
avatar use. However, there were very few negative events that happened to the
avatars in this scenario. As mentioned earlier, participants won 90% of the rounds
they played. Thus, participants had little opportunity to experience negative
emotional valence while using their avatars.
The lack of a relationship between any level of self-presence and any measure
of arousal illustrates a limitation in the present analysis that is rooted in the
physiological measurement of arousal. Namely, HR and EDA are much more
sensitive to movement than psychological influences (Hugdahl, 1996). Given that the
game involved somewhat vigorous physical activity (i.e., swinging a virtual sword
through the air), the resulting variance in HR and EDA was likely so large that it

103
would have eclipsed any changes related to self-presence. Although the interaction
effect between avatar customization and gender-consistency did influence EDA, the
strength of this psychological influence is justifiably stronger because it relates to
discomfort and aversion. In other words, people are probably more easily induced to
feel upset than to feel self-presence.
Overall, these results provide notable insights into physiological responses
during avatar use, but the current design limits any claims about the validity of self-
presence and the SPQ. These limitations result mostly from the concurrence of the
physiological and self-presence experiences, the lack of opportunities for negative
events during avatar use, and the difficulty of measuring arousal during movement.
Therefore, a time order-controlled test of the association between self-presence and
physiological responses with negative events directed toward the avatars, and minimal
participant movement would provide greater insights into the validity of self-presence
and the SPQ. The following chapter provides such a test.

104
Chapter 6: Physiological Effects of Self-Presence after Avatar Use

Chapter 6: Abstract
The present chapter examines self-presence within the same experimental
context described in the previous two chapters, but focuses on the direct and causal
influence of self-presence on physiological responses. The robust test for this
relationship includes physiological measurement after initial avatar use while
participants watched, without controlling, their avatars in a context where the avatars
received negative treatment. Proto self-presence during avatar use was found to
negatively influence arousal and negative emotional responses while subsequently
watching the avatar, indicating that proto self-presence ceases immediately after
disconnection from the avatar and thus people who experience more proto self-
presence during avatar use are more affected by this disconnection. Core and
extended self-presence during avatar use were found to positively influence
subsequent arousal and negative emotional responses, respectively, indicating that
these levels of self-presence persist after disconnection from the avatar. These results
are consistent with the conceptual framework of self-presence and suggest that this
line of research should shift focus from testing the framework of self-presence itself
to using it in examinations of other phenomena, such as learning in video games.

Differences in self-presence may cause differences in physiological responses,
but due to the measurement order issue noted in the previous chapter, such causal
relationships are not testable during avatar use. However, they are testable after
initial avatar use, during subsequent engagement with the avatar. If differences in
physiological responses during such subsequent avatar engagement correspond with

105
differences in responses to the Self-Presence Questionnaire (SPQ), then the
relationship is likely causal, given that potential common causes of self-presence and
the subsequent physiological responses are taken into account. For example,
physiological responses during avatar use may influence responses to the SPQ after
avatar use, as well as physiological responses during a subsequent avatar use. By
controlling for this latter relationship (physiological responses during avatar use
affecting physiological responses afterward), any relationship discovered between
self-presence and physiological responses after initial avatar use is most likely causal.
As indicated in the previous chapter, physical movement during avatar use
creates noise in physiological output. Restricting movement during this subsequent
engagement would facilitate cleaner comparisons of self-presence and physiological
indicators. In other words, after establishing self-presence during initial avatar use,
people’s physiological responses could be measured while they watch but not control
their avatars. Their experience of self-presence during initial avatar use may
influence these responses, just as observing but not controlling an owned avatar
affects the owner’s responses (Fox & Bailenson, 2009; Fox, et al., 2009).
Also as noted in the previous chapter, game contexts that include mostly
positive interactions between the avatar and virtual environment do not provide the
opportunity to compare the relationship between self-presence and negative emotional
valence. After initial use, while participants watch the avatar, if the virtual
environment were to treat the avatar in a negative manner, this would facilitate an
examination of the negative emotional responses (or reduced positive responses) and
self-presence. Further, because physiological responses tend to be higher for negative
stimuli (Courtney, et al., 2010; Lovibond, 1992), such a context would influence
psychological arousal more strongly than a positive-treatment context.

106
Together, this reasoning suggests that a robust test for causal relationships
between self-presence and physiological responses should include physiological
measurement after initial avatar use and response to the SPQ, while participants watch
but do not control their avatars in a context where the avatars receive negative
treatment. In order to isolate the effects of self-presence on physiological responses
within such a context, other potentially influential factors present during the avatar
use context must be taken into account. For example, avatar-assignment condition
influenced self-presence and physiological responses during avatar use and thus may
distinctly affect physiological responses after avatar use, while participants watch
their avatars receive negative treatment. Given that the purpose of the present chapter
is not to focus on such relationships, these variables are treated only as covariates.
Instead, this chapter focuses on the systematic and distinct effects of the three levels
of self-presence on physiological responses, thereby providing unique validity tests of
self-presence and the SPQ.
Proto self-presence is conceptualized to occur only during avatar use (Chapter
1), so when watching but not using the avatar, people are not likely to feel this body-
level connection to the avatar. In other words, just as the integration of false body
parts into body schema is hindered by changing the parts’ orientation (Armel &
Ramachandran, 2003; Tsakiris & Haggard, 2005), restricting someone from using the
avatar control device is likely to inhibit proto self-presence even if the avatar is
visible. However, the recent experience of proto self-presence may still cause
differences in responses while watching the avatar because the disconnection from the
avatar would feel more salient to people who experienced more proto self-presence.
The more salient this disconnection, the less likely the person is to focus on the avatar
and respond physiologically after disconnection. Thus, people who report higher

107
levels of proto self-presence are expected to exhibit weaker physiological responses
after avatar use, while watching the avatar receive negative treatment.
H1. Proto self-presence during avatar use negatively influences physiological arousal
after avatar use, while the avatar receives negative treatment.
H2. Proto self-presence during avatar use negatively influences negative emotional
responses after avatar use, while the avatar receives negative treatment.
In contrast, core self-presence can persist for minutes or hours after avatar use
(Chapter 1). Someone who experiences a higher degree of core self-presence during
avatar use is likely to feel more of this emotion-level connection afterwards and thus
exhibit stronger physiological responses when watching the avatar receive negative
treatment. Thus, core self-presence is expected to influence physiological output in
the opposite way from proto self-presence.
H3. Core self-presence during avatar use positively influences physiological arousal
after avatar use, while the avatar receives negative treatment.
H4. Core self-presence during avatar positively influences negative emotional
responses after avatar use, while the avatar receives negative treatment.
Extended self-presence also persists after avatar use (Chapter 1), and thus it
should influence physiological output in the same way as core self-presence. People
who experience this identity-level connection to the avatar may dislike observing their
avatars receive negative treatment and thus exhibit stronger physiological responses.
This would be consistent with the finding that people who were given a choice of
their avatars exhibited more arousal (EDA) while playing a game with the avatar than
people who were assigned to use a specific avatar (Lim & Reeves, 2009).
H5. Extended self-presence during avatar use positively influences physiological
arousal after avatar use, while the avatar receives negative treatment.

108
H6. Extended self-presence during avatar use positively influences negative emotional
responses after avatar use, while the avatar receives negative treatment.
Method
The 2x2 experiment described in the previous two chapters is the same one
used to examine the present hypotheses. The previous methods sections have
described most of this experiment in detail, so the present description only provides
details on the procedures and measures that are relevant to the hypotheses presented
in this chapter.
Procedure
As described in the previous two chapters, the participants entered the lab,
read the IRB-approved information sheet, completed the initial survey, designed an
avatar (gender assigned according to condition), had the physiological recording
sensors attached to their bodies, relaxed during one minute of baseline recording,
played a sword-fighting game (avatar assigned according to condition) for
approximately 10 minutes, and then completed a 15-minute survey. The segment of
the experiment immediately after this survey is the main focus of the present chapter.
First, participants were instructed to use the Wii controller to click through the menu
options and prepare for another session of play. Participants chose the same avatar
they had used in the first play session, and instead of the regular 1-on-1 Swordplay
game, they selected the “Showdown” sub-game of Swordplay. This sub-game is
made for single-players to battle a series of computer-controlled opponents.
However, before the game began, the researcher instructed the participants to put the
controller down on the couch and just watch the game. The participants then watched
as their avatars were defeated after getting hit 3 times by the computer-controlled
opponent. In addition to calculating the average physiological output during this

109
segment, the researcher marked the moment of each of these 3 hits so that event-
related physiological responses could be analyzed. The amount of time between hits
varied randomly, thus inducing variance in the total duration of this segment of the
experiment (m = 38.00 seconds, sd = 14.13). The researcher then turned off the
television and administered the final questionnaire.
Measures and Metrics
The physiological indicators in the present chapter – heart rate (HR),
electrodermal activity (EDA), zygomatic (cheek) electromyogram (EMG), and
corrugator EMG – are the same as described in the previous chapter. However, some
of the measures used for this analysis differ. In addition to tonic comparisons of
average physiological values over time, as used in the previous chapter, the present
analysis includes phasic measures, which allow for comparisons of discrete responses
to specific events (Hugdahl, 1996).
As in the previous chapter, the present tonic measures are based on averages
of each type of physiological output over the entire session of watching the avatar.
The specific transformations used in the previous chapter for each type of
physiological measure were retained in the present analysis. Thus, the tonic HR
metric was standardized into a z score (Ben-Shakhar, 1985): (value – mean) / standard
deviation. The tonic EDA metric was based on ranged correction (Lykken, et al.,
1966): (value – minimum) / (maximum – minimum). Tonic facial EMG metrics were
calculated as percent change from baseline (Kaiser, 1989): (value – baseline) /
baseline.
The phasic measures are based on discrete changes associated with each of the
three times the participants’ avatars were hit by the opponent. For HR, this measure
was constructed by taking an average over the three-second period prior to the hit and

110
the three-second period following the hit and then subtracting the latter from the
former. It was hypothesized that HR would decrease after the hit. Although
somewhat counterintuitive, such HR deceleration reflects an orienting response to
novel stimuli and has been positively associated with arousal (Hugdahl, 1996). For
EDA, this measure was constructed by subtracting the minimum from the maximum
value during the three-second period following the hit. It should be noted that this
measure was only calculated for true responses. Thus, if EDA decreased after a hit,
this was counted as a non-response. Finally, for the facial EMG measures, the phasic
measures were constructed by subtracting the minimum from the maximum values
during the three-second period following the hit. Based on an analysis of dependence
on baseline (Kaiser, 1989), as conducted in the previous chapter, the metrics derived
from these phasic measures are based on percentage change from baseline rather than
absolute changes..
The present analysis examines the three hits separately because order is likely
to play a role in responses. The first hit was novel to the participant and thus most
surprising. After the first hit, participants were aware that their avatar would probably
be hit again. Thus, they were likely to pay more attention (orient) toward the game
environment. After the second hit, there was less uncertainty that the avatar would be
hit again. But following the third hit, the animation of the avatar flying backwards
was played in a dramatic slow motion, indicating that the avatar had been completely
defeated. In summary, the first hit was most surprising, the second attracted the most
attention, and the third was the most dramatic. Because of these differences, responses
to each hit were analyzed individually, though the distinctions between the hits were
not considered in the formulation of the hypotheses.

111
As described earlier, in order to isolate the causal relationship between self-
presence and physiological responses while watching the avatar receive negative
treatment, the participants’ physiological state during avatar use must be taken into
account. Thus, the physiological metrics used as dependent variables in the previous
analysis are treated as covariates in the present one. Also as described above, the
avatar-assignment conditions– avatar customization, gender consistency, and the
interaction between the two – are not examined in the present analysis and thus are
also treated as covariates. Finally, the present analyses include all of the other
covariates used in the previous analyses to control for potential effects as described in
the previous chapter. Specifically, these covariates include measures of presence,
frequency of video game playing, percentage of rounds won, and participant race.
Chapter 6: Results

In order to test the present hypotheses, each metric constructed from the
physiological measures taken while participants watched their avatars receive
negative treatment was entered as the dependent variable in a series of regression
analyses. For the tonic metrics, as in the previous chapter, this resulted in four
separate regression models. For the phasic metrics, however, there were twelve
separate models because responses to each of the three hits were examined for each of
the four physiological measures. In order to simplify presentation, the results for the
tonic analyses are presented with two dependent variables in each table. For example,
Table 6.1 contains the regressions with the tonic HR dependent variable as well as the
tonic EDA dependent variable. Further, for the phasic analyses, results for each
dependent variable are presented for all three hits within individual tables. For
example, Table 6.3 contains the regressions for hits 1, 2 and 3, with the phasic HR
metric as the dependent variable. All results are presented in Tables 6.1-6.6 below.

112
Table 6.1. Regression Models on Tonic Arousal Measures

DV: HR
DV: EDA

Cohen's

Cohen's
Beta t d size^ Beta t d size^
Proto self-presence -0.03 -0.16

0.13 0.68

Core self-presence -0.07 -0.47

-0.01 -0.05

Extended self-
presence -0.14 -0.91

-0.05 -0.29

Avatar customization
0.20 1.25

0.16 1.03

Gender consistency
-0.07 -0.55

0.10 0.75

Interaction (custom *
gender)
-0.24 -1.82
†
-0.52 med -0.11 -0.71

Presence 0.30 1.69
†
0.48 med 0.01 0.05

Race (white)
0.17 1.27

0.27 1.94
†
0.55 med
Video game
frequency -0.08 -0.64

0.11 0.80

Percent win -0.09 -0.75

-0.04 -0.33

In-game Physio
(HR/EDA) -0.40 -3.14
**
-0.90 large 0.33 2.31
*
0.65 med

df 49 50
R
2

0.32

0.14

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2 small

113
Table 6.2. Regression Models on Tonic Emotional Valence Measures

DV: Zygomatic EMG
(cheek)
DV: Corrugator EMG
(forehead)

Cohen's

Cohen's
Beta t d size^ Beta t d size^
Proto self-presence 0.16 0.76

-0.49 -2.87
**
-0.84 large
Core self-presence -0.05 -0.30

-0.09 -0.63

Extended self-presence -0.10 -0.57

-0.12 -0.82

Avatar customization
0.19 1.09

0.02 0.12

Gender consistency
0.13 0.88

0.04 0.32

Interaction (custom *
gender)
0.03 0.19

0.26 2.00 p=.05 0.58 med
Presence -0.04 -0.22

0.44 2.80
**
0.82 large
Race (white)
-0.15 -1.05

0.17 1.42

Video game frequency 0.03 0.21

0.13 1.03

Percent win 0.01 0.08

0.19 1.61

In-Game Physio
(Zygo/Corr) 0.32 2.14
*
0.62 med 0.41 3.16
**
0.92 large

df 48 47
R
2

0.20

0.42

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2 small

114
Table 6.3. Regression Models on the Phasic HR Metric for Hits 1-3

Hit 1
Hit 2

Cohen's
Cohen's
Beta t d size^ Beta t d size^
Proto self-presence 0.10 0.47

-0.40 -2.06
*
-0.60 med
Core self-presence -0.17 -0.94

0.39 2.28
*
0.66 med
Extended self-presence 0.16 0.91

-0.21 -1.26

Avatar customization
0.06 0.32

0.37 2.17
*
0.63 med
Gender consistency
-0.17 -1.11

0.02 0.16

Interaction (custom *
gender)
-0.06 -0.39

0.13 0.87

Presence 0.07 0.38

0.32 1.71
†
0.50 med
Race (white)
-0.02 -0.15

0.29 2.05
*
0.60 med
Video game frequency -0.14 -0.94

0.02 0.13

Percent win 0.12 0.82

0.18 1.35

In-Game HR -0.26 -1.74
†
-0.51 med -0.04 -0.26

df 47 47
R
2

0.14

0.24

Hit 3

Cohen's
Beta t d size^
Proto self-presence 0.13 0.60

Core self-presence -0.06 -0.33

Extended self-presence 0.07 0.38

Avatar customization
-0.21 -1.12

Gender consistency
0.10 0.65

Interaction (custom *
gender)
-0.03 -0.20

Presence -0.08 -0.38

Race (white)
-0.18 -1.18

Video game frequency 0.09 0.63

Percent win 0.02 0.15

In-Game HR 0.23 1.58

df 47
R
2
0.11

^ >0.8 considered large, 0.2-0.8 medium, <.2
small

115
Table 6.4. Regression Models on the Phasic EDA Metric for Hits 1-3

Hit 1
Hit 2

Cohen's

Cohen's
Beta t d size^ Beta t d size^
Proto self-presence -0.20 -0.77

0.08 0.19

Core self-presence -0.09 -0.42

-0.55 -1.00

Extended self-presence -0.44 -2.07
*
-0.81 large 0.02 0.04

Avatar customization
0.09 0.50

-0.47 -0.86

Gender consistency
0.03 0.17

-0.13 -0.38

Interaction (custom *
gender)
0.43 2.42
*
0.95 large 0.55 1.27

Presence 0.66 2.67
*
1.05 large 0.14 0.35

Race (white)
0.11 0.55

-0.13 -0.45

Video game frequency -0.31 -1.73
†
-0.68 med -0.43 -0.99

Percent win -0.05 -0.27

-0.15 -0.50

In-Game EDA 0.31 1.75
†
0.68 med 0.77 2.20
†
1.47

df 26 9
R
2

0.44

0.50

Hit 3

Cohen's

Beta t d size^
Proto self-presence -0.30 -0.78

Core self-presence 0.44 1.22

Extended self-presence -0.74 -2.10
†
-1.16 large
Avatar customization
-0.30 -1.29

Gender consistency
0.03 0.11

Interaction (custom *
gender)
0.56 1.50

Presence 0.33 1.18

Race (white)
0.41 1.67

Video game frequency 0.19 0.68

Percent win -0.24 -1.06

In-Game EDA -0.17 -0.60

df 13

R
2
0.50

^ >0.8 considered large, 0.2-0.8 medium, <.2
small

116
Table 6.5. Regression Models on the Phasic Zygomatic EMG Metric for Hits 1-3

Hit 1
Hit 2

Cohen's

Cohen's
Beta t d size^ Beta t d size^
Proto self-presence -0.22 -1.15

-0.13 -0.68

Core self-presence -0.07 -0.41

0.13 0.80

Extended self-presence 0.04 0.23

-0.04 -0.26

Avatar customization
-0.01 -0.06

0.05 0.29

Gender consistency
0.27 1.96
†
0.55 med 0.24 1.75
†
0.49 med
Interaction (custom *
gender)
-0.13 -0.92

-0.09 -0.65

Presence 0.33 1.81

0.25 1.38

Race (white)
-0.15 -1.09

-0.24 -1.78
†
-0.50 med
Video game frequency 0.13 0.91

-0.03 -0.25

Percent win -0.06 -0.47

-0.07 -0.52

In-Game Zygomatic EMG 0.40 2.86
**
0.81 large 0.38 2.79
**
0.79 med

df 50 50
R
2

0.25

0.27

Hit 3

Cohen's

Beta t d size^
Proto self-presence 0.08 0.44

Core self-presence 0.23 1.60

Extended self-presence -0.31 -2.15
*
-0.61 med
Avatar customization
0.23 1.61

Gender consistency
0.36 3.00
**
0.85 large
Interaction (custom *
gender)
-0.03 -0.21

Presence 0.19 1.21

Race (white)
-0.17 -1.42

Video game frequency -0.08 -0.69

Percent win 0.19 1.68

In-Game Zygomatic EMG 0.35 2.87
**
0.81 large

df 50

R
2
0.43

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2 small

117

Table 6.6. Regression Models on the Phasic Corrugator EMG Metric for Hits 1-3

Hit 1

Hit 2

Cohen's
Cohen's
Beta t d size^ Beta t d size^
Proto self-presence -0.24 -1.23

-0.22 -1.12

Core self-presence -0.23 -1.34

-0.29 -1.70

Extended self-presence 0.12 0.71

-0.06 -0.34

Avatar customization
-0.09 -0.53

-0.13 -0.80

Gender consistency
0.23 1.62

0.02 0.14

Interaction (cust * gender)
0.05 0.30

0.03 0.23

Presence 0.23 1.26

0.32 1.77
†
0.51 med
Race (white)
-0.14 -1.01

0.06 0.45

Video game frequency 0.03 0.22

-0.09 -0.64

Percent win 0.07 0.50

0.22 1.67

In-Game Corrugator EMG 0.32 2.15
*
0.61 med 0.21 1.43

df 49 49
R
2

0.21

0.24

Hit 3

Cohen's
Beta t d size^
Proto self-presence -0.44 -2.40
*
-0.68 med
Core self-presence 0.23 1.45

Extended self-presence 0.05 0.33

Avatar customization
0.01 0.06

Gender consistency
0.15 1.15

Interaction (cust * gender)
-0.09 -0.61

Presence 0.31 1.83
†
0.52 med
Race (white)
0.16 1.25

Video game frequency -0.08 -0.61

Percent win 0.10 0.79

In-Game Corrugator EMG 0.43 3.13
**
0.90 large

df 49
R
2
0.33

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2 small

118
These results provide mixed support for the hypotheses. Hypothesis 1, which
posits that proto self-presence negatively influences subsequent arousal, was
supported by phasic HR change for hit 2, as illustrated in Table 6.3. The higher the
proto self-presence, the lower the change in HR after hit 2, and thus the lower the
arousal while watching the avatar receive negative treatment. Hypothesis 2, which
posits that proto self-presence negatively influences subsequent negative emotional
responses, was supported by tonic corrugator EMG activity as well as phasic
corrugator EMG activity after hit 3, as illustrated in Tables 6.2 and 6.6, respectively.
The higher the proto self-presence, the lower the corrugator EMG activity (frowning),
and thus the lower negative emotional responses while watching the avatar receive
negative treatment.
Hypothesis 3, which posits that core self-presence positively influences
subsequent arousal, was supported by phasic HR change after hit 2, as illustrated in
Table 6.3. The higher the core self-presence, the higher the change in HR after hit 2,
and thus the higher the arousal while watching the avatar receive negative treatment.
Hypothesis 4, which posits that core self-presence positively influences subsequent
negative emotional responses, was not supported by any of the results.
Hypothesis 5, which posits that extended self-presence positively influences
subsequent arousal, was not supported, but phasic EDA responses to hits 1 and 3
support the opposite direction of this relationship, as illustrated in Table 6.4. The
higher the extended self-presence, the lower the EDA responses after hits 1 and 3, and
thus the lower the arousal while watching the avatar receive negative treatment. This
finding should be treated cautiously because of the low degrees of freedom in the
regression, which resulted from a high non-response rate for phasic EDA responses
after each hit. Hypothesis 6, which posits that extended self-presence positively

119
influences subsequent negative emotional responses, was supported by phasic
zygomatic EMG activity after hit 3, as illustrated in Table 6.5. The higher the
extended self-presence, the lower the zygomatic EMG activity (smiling) after hit 3,
and thus the greater negative emotional responses while watching the avatar receive
negative treatment.
Chapter 6: Discussion

The present chapter examines whether self-presence influences physiological
responses during subsequent avatar engagement. The experiment was designed to
isolate this causal relationship by having participants watch but not control their
avatars while these avatars received negative treatment. Physiological responses were
consistent with many of the expectations. The extent of proto self-presence during
avatar use negatively influenced arousal and negative emotional responses while
subsequently watching the avatar. The extent of core self-presence during avatar use
positively influenced subsequent arousal. And the extent of extended self-presence
negatively influenced subsequent positive emotional responses, which is tantamount
to positively influencing negative emotional responses. Unexpectedly, extended self-
presence was found to negatively influence subsequent arousal, as indicated by a
measure of phasic EDA, but this result is dubious because of a high non-response rate
for this measure. Overall, these results indicate that the three levels of self-presence
distinctly affect responses to avatars in ways that are mostly predicable and reflected
by physiological indicators. This is the strongest support for the construct validity of
self-presence and the Self-Presence Questionnaire (SPQ) measurement tool presented
in this dissertation.
The negative relationship between proto self-presence during avatar use and
physiological responses after being disconnected from but watching the avatar

120
supports the reasoning that such disconnection is more salient for people who
experience more proto self-presence. Thus, these people experience less proto self-
presence while watching the avatar than the people who experienced less proto self-
presence while using the avatar. The complement to this relationship is that the
people who experience less proto self-presence during avatar use are the people who
experience more proto self-presence while watching the avatar. In other words, the
relative difference in proto self-presence during avatar use switches after
disconnection from the avatar. The most notable implication of this finding is that
proto self-presence causes stronger physiological arousal and emotional responses,
thus supporting the construct validity of this level of self-presence. In other words,
the more proto self-presence someone experiences, the more the avatar is treated as an
extension of body schema and thus the more the individual responds to use of the
avatar as it were part of the individual’s body.
The positive relationship between core self-presence during avatar use and HR
change while watching the avatar supports the reasoning that core self-presence
persists after disconnection from the avatar. Further, the HR decrease associated with
increased core self-presence indicates not just increased arousal, but also attention
allocation. Although the two are interrelated, this suggests that future research
utilizing core self-presence should attempt to distinguish whether it is related equally
to arousal and attention. Regardless, this test indicates that core self-presence
increases the extent to which avatar users feel an emotion-level connection to the
avatar, which then increases physiological responses to interactions between the
avatar and the virtual environment.
The positive relationship between extended self-presence during avatar use
and reduced zygomatic EMG (smiling) while watching the avatar receive negative

121
treatment supports the reasoning that this identity-level connection to an avatar
persists after disconnection from the avatar. Further, the experience of extended self-
presence causes an increase in the extent to which avatar users dislike watching their
avatars receive negative treatment, thus inhibiting positive emotional responses, i.e.,
smiling. Given that avatar customization was found to increase extended self-
presence (Chapter 4), this finding is consistent with the finding that having a choice of
avatar (compared to assignment) increases physiological responses during avatar use
(Lim & Reeves, 2009).
The finding that extended self-presence was negatively related to
physiological arousal (EDA) contradicts this finding. As mentioned earlier, this
contradiction may result from a statistical anomaly caused by the large non-response
rate for EDA while watching the avatars receive negative treatment. An analysis of
the full regression model for this effect (phasic EDA responses to Hit 1) indicates that
this is the case. Specifically, the descriptive statistics (Table 6.7) for the significant
interaction effect of avatar-assignment condition on this dependent variable illustrate
that the high mean for the generic, opposite gender condition drives this effect. The
lower N and higher SD for this condition suggest that a small portion of participants
drive this mean. Thus, the relationship between extended self-presence and decreased
physiological arousal while watching the avatar receive negative treatment may be
due to chance and should be treated cautiously.

122
Table 6.7. Means of Phasic EDA Change after Hit 1 by Avatar Assignment Condition
Condition Mean N SD
Customized, same gender 0.16 10 0.13
Customized, opposite gender 0.11 11 0.10
Generic, same gender 0.13 10 0.18
Generic, opposite gender 0.23 7 0.30
Total 0.15 38 0.18

Two other unsupported hypotheses that merit attention. Namely, core and
extended self-presence were not found to influence emotional valence and arousal,
respectively. Although core self-presence was found to increase arousal, it may be
the case that this emotional-level connection to the avatar relates mostly to the
autonomic nervous system’s “fight or flight” response and not to a level of emotional
responses that incorporates valenced value judgments. Another possibility is that the
items included in the current composite measure of core self-presence (“sad”, “scary”,
and “aroused”), which were influenced by the specific virtual environment, contain a
much greater proportion of arousal-related as opposed to valence-related adjectives.
Though the “sad” and “scary” items are negatively valenced, the latter also relates to
arousal. This suggests that the relationship between core self-presence and emotional
valence may depend on the type of virtual environment in which avatar use occurs.
Similarly, the lack of a relationship between extended self-presence and arousal in the
current study is likely due to specific facets of the virtual environment or study
design. Unlike in Lim and Reeves (2009) study, where half of the participants chose
an avatar and then exhibited higher EDA when using the avatar, all participants in the
present study customized an avatar, but only half of them used the avatar they
customized. Thus, the participants who used generic avatars not only experienced
less extended self-presence but were also disappointed by the inability to use an
avatar they had customized. The feeling of disappointment may have persisted

123
throughout the study experience, and given that people who experience aversive
stimuli-induced stress exhibit stronger EDA responses (Courtney, et al., 2010;
Lovibond, 1992), this disappointment may have increased arousal when watching the
avatars receive negative treatment. In other words, any reductions in EDA responses
caused by a lack of extended self-presence may have been counteracted by increases
in EDA responses caused by the disappointment of using a generic avatar. This
confound was included in the present study in order to ensure that all participants
experienced the same order of operations, but future studies could be designed to
avoid such a confound.
Another point worth mentioning is the difference in results between the three
hits. To generalize, Hit 1 was most surprising and most strongly related to EDA
effects, Hit 2 attracted the most attention and was most strongly related to HR effect,
and Hit 3 was most dramatic and most strongly related to Zygomatic and Corrugator
EMG effects. This is consistent with the formulation of the substantive differences
between the hits because EDA responses are associated with surprise, HR responses
are associated with attention, and facial EMG responses are associated with emotional
responses (Hugdahl, 1996). Although this explanation for the differences in results
between the hits is parsimonious, it is not ideal because the experiment was not
designed to hypothesis test for these differences. Future research should either take
such order effects into account more formally or should isolate order altogether.
Overall, the results presented in the current chapter indicate that the three
levels of self-presence distinctly affect physiological responses to avatars in ways that
are consistent with expectations about how self-presence influences psychological
state. This is the strongest support for the construct validity of self-presence and the
SPQ measurement tool presented in this dissertation. While there are still some open

124
validity tests to which future research should contribute, the current accumulation of
support suggests that the main focus of this line of research should shift from testing
the concept of self-presence itself to using the concept to understand other
phenomena. The following chapter presents such an application.

125
Chapter 7: Self-Presence and Math Performance

Chapter 7: Abstract
The present chapter provides the first empirical application of this
dissertation’s framework of self-presence to a phenomenon besides self-presence
itself. Specifically, this chapter examines how avatar use may influence learning in
video games and the role that self-presence may play in this relationship. Drawing
from results on a math task during the experiment described in the previous three
chapters, the initial intent of this examination was to test whether self-presence plays
a role in the stereotype threat-induced relationship between avatar gender and math
performance. However, the stereotype threat prompt used in this study was not
threatening enough to produce such a relationship, so the results do not contribute to
this original goal. Still, the result that proto self-presence was negatively related to
performance on the math task has notable potential implications. This effect may be
due to the increased difficulty in changing cognitive processing style from embodied
to abstract for people who experience more proto self-presence. It may also be due to
an individual trait that is common to both experiencing proto self-presence and
performing poorly on math tasks. Or it may be due to the positive relationship
between proto self-presence and enjoyment, the latter of which may hinder
subsequent focus. Regardless of this specific interpretation, the present chapter
represents an important first application of self-presence to avatar use effects research.

As described in Chapter 1, the concept of self-presence may contribute to our
understanding of a variety of phenomena that occur during or after avatar use. While
research in this growing area often occurs in controlled environments (e.g., Fox, et al.,

126
2009; Groom, Bailenson, et al., 2009; Jin, 2009; Yee, et al., 2009), the ultimate goal
of such research is to develop an understanding of the ways that everyday media use
affects people and society. To this end, the present chapter applies the concept of
self-presence to a common type of media that has the potential to improve lives and
society: learning-oriented video games.
Although video games have received negative attention because of their
potential relationship with aggression or addiction (Anderson, 2002; Bushman &
Huesmann, 2006; Ng & Wiemer-Hastings, 2005; Salguero & Moran, 2002; Weber, et
al., 2006), there is a growing body of games with positive goals as well as a growing
body of research about them. The positive foci of these games range vastly, from
improving academic learning to inducing social change, though the former category
has historically been the largest (Ratan & Ritterfeld, 2009). Some games contribute to
occupational training in high-skill areas, such as airplane piloting or laparoscopic
surgery (Hart & Battiste, 1992; Rosser Jr, et al., 2007). Others focus on a wide range
of health improvement goals (Gamberini, Barresi, Majer, & Scarpetta, 2008). Some
of these games aim to increase knowledge about and management of health
conditions, such as asthma and cancer (Bartholomew, et al., 2000; Beale, Kato,
Marin-Bowling, Guthrie, & Cole, 2007; Kato, Cole, Bradlyn, & Pollock, 2008).
Others focus on physical rehabilitation, such as for patients recovering from cerebral
palsy or stroke (Alankus, Lazar, May, & Kelleher; Deutsch, Borbely, Filler, Huhn, &
Guarrera-Bowlby, 2008). Some games aim to improve the general population’s
health by motivating healthy eating and exercise behaviors (Baranowski, et al., 2003;
Warburton, et al., 2007). Others have been found to improve cognitive functions,
such as vision and spatial cognition (Feng, Spence, & Pratt, 2007; Green & Bavelier,
2003; Li, Polat, Makous, & Bavelier, 2009) as well as the reduction of cognitive

127
decline in older adults (Basak, Boot, Voss, & Kramer, 2008; Nacke, Nacke, &
Lindley, 2009). This plethora of examples suggests that video games can contribute
positively to our society and that such contributions would be likely to increase as the
number and quality of positive games and research on such games increases.
The present chapter uses the concept of self-presence to examine the ways in
which an understanding of virtual self-representations, i.e., avatars, can lead to video
game uses that generate better learning outcomes. Just as avatar attributes have been
found to affect social, economic, and health-related behaviors in virtual environments
(Fox & Bailenson, 2009; Groom, Bailenson, et al., 2009; Yee & Bailenson, 2006;
Yee, et al., 2009), certain types of avatar use may also influence learning in such
environments. For example, self-presence may play a role in the recent findings that
the extent to which a virtual training agent resembles the user, the more the user
engages with and learns from the trainer (Behrend & Thompson, in press, 2011).
Games provide a suitable context in which to promote learning because they induce a
state of flow, thereby making tasks more exciting and motivational (Nacke, et al.,
2009; Sherry, 2004). Self-presence was found to influence physiological arousal
within a gaming context (Chapter 6) and thus it may also influence attention and
motivation (Lang, et al., 2009) within learning-oriented games. In other words,
because the ways and extent to which people connect to their avatars in learning-
oriented games have the potential to improve or hinder learning-related outcomes, the
concept of self-presence may contribute to an understanding of such processes.
Stereotype Threat and Avatars
There has been little research on avatar-related learning effects, but Lee’s
(2009) recent dissertation presents three studies in which avatar gender influenced
participants’ performance on a math test. This effect is explained by the phenomenon

128
of stereotype threat (Steele, 1997; Steele & Aronson, 1995), namely, that reminders of
identity-associated stereotypes induce conformity to those stereotypes, thereby
reducing performance and motivation for people who associate with the stereotyped
identity. Specifically, in Lee’s (2009) research, people who used female avatars in
majority male groups performed worse than people who used male avatars when
placed in a context that emphasized the competition between males and females. In
other words, when reminded (implicitly) of the negative stereotype that females tend
to perform worse at math (Steele, 1997; Steele & Aronson, 1995), participants who
used female avatars were hindered in performance and motivation on the math task
(Lee, 2009). This finding is especially notable because in one of Lee’s studies avatar
gender influenced math performance regardless of the participants’ biological sex.
The implicit reminder of the stereotype was administered via gender
distribution and competitive context. Specifically, the stereotype threat effect
occurred when the avatar was portrayed in the minority gender (one female and two
males in the group) and the context of the math task as described as competitive
(females versus males), as opposed to cooperative (all group members together).
These conditions induced the participants to feel as though the avatar’s gender
identity was relevant to their own identities. Although Lee did not use the
terminology, these conditions essentially induced extended self-presence. Thus,
stereotype threat influenced performance on the math task for those participants who
were induced to experience a high degree of extended self-presence.
While performance on a math task is not a measure of learning itself, it
indicates the extent to which the participants are able and motivated to engage
cognitive resources. Because such ability and motivation are required for learning,
performance on the task reflects the ability to learn in a knowledge-acquisition

129
context. Thus, this finding implies that avatars in learning-oriented games have the
potential to induce stereotype threat and thus harm the learning experience when the
experience of extended self-presence is high.
The present chapter aims to empirically test this relationship as a means of
determining how self-presence may influence performance in learning-oriented
games. But in order to examine the relationship between avatar gender, self-presence,
and math performance, the main effect between avatar gender and math performance
must be replicated. The following hypothesis articulates this expectation.
Hypothesis 1: After receiving a stereotype threat prompt, participants who use a
female avatar will perform worse at a math task than participants who use a male
avatar.
As described above, extended self-presence should augment this proposed
relationship between avatar gender and math performance. The more the avatar is
perceived as an extension of the self at the identity level, the more the stereotype
threat prompt is likely to affect someone using a female avatar. The following
hypothesis reflects this expectation.
Hypothesis 2: The relationship between avatar gender and math task performance is
stronger for people who experience more extended self-presence.
Both core and proto self-presence may contribute to this relationship as well,
but the direction of influence is unclear. On the one hand, given that the three levels
of self-presence are positively related (Chapters 2-4), these emotion and body-level
connections to the avatar may reinforce the identity-connection to the avatar and thus
influence the relationship in the same direction as proposed in Hypothesis 2. On the
other hand, just as proto self-presence was found to affect physiological responses
after disconnection from the avatar in the opposite direction as core and extended self-

130
presence, the extent of proto self-presence may negatively influence the strength of
the stereotype threat effect because this effect also happens after disconnection from
the avatar. Similarly, because core self-presence is conceptualized to last for a shorter
amount of time after disconnection from the avatar than extended self-presence
(Chapter 1), the effect of the stereotype threat prompt may not be influenced by core
self-presence. Thus, the potential relationship between these constructs is examined
within the context of a research question.
Research Question 1: Is the relationship between avatar gender and math task
performance influenced by core or proto self-presence?
Chapter 7: Method
The same study as used in the Chapters 4-6 was utilized to test the hypotheses
and research question. As in Chapters 5-6, the full procedure is not described here, but
instead only those aspects of the study that are relevant to the present hypotheses are
discussed.
Procedure
As described in the previous chapters, the participants entered the lab, read the
IRB-approved information sheet, completed the initial survey, designed an avatar
(gender assigned according to condition), had the physiological recording equipment
attached to their bodies, relaxed during one minute of baseline recording, played a
sword-fighting game (avatar assigned according to condition) for approximately 10
minutes, and then completed a 15-minute survey. The present chapter focuses on the
results from the final 5 minutes of this survey. About 10 minutes into the survey, the
participants reached a page that said, “Please stop and tell the researcher you have
reached this page.” The researcher then prompted them to click next and read the
instructions, “You will now have 5 minutes to complete some challenging math

131
questions in your head. The scores of people who used MALE Miis will be
COMPARED to the scores of people who used FEMALE Miis.
5
” The researcher then
confirmed that the participants understood that people who used male avatars would
have their scores compared to people who used female avatars. The participants then
completed the 10 mental arithmetic questions on the ipad. After five minutes, if the
participants were not done with the questions, the researcher told them to stop
working and click to the next page. The remainder of the study progressed and
concluded as described in Chapter 6. The final question on the final questionnaire
asked participants to report their math SAT score.
Measures
The present analysis includes fewer measures than in most of the previous
chapters. The 10 mental math questions, taken directly from the 20 questions
included in Lee’s (2009) examination of the same topic, were chosen for their
difficulty level and can be found in the Chapter Appendix. It should be noted that 10
instead of 20 questions were used because the present study only allowed for a five-
minute math response segment while Lee’s took 10 minutes. The metric used for the
analysis was the same as Lee used, number of questions that the participant answered
correctly (mean = 6.44, sd = 2.05). The same composite measures of the three levels
of self-presence described in Chapter 4 were used in the present analysis. As in Lee’s
(2009) studies, SAT score was included as a covariate to control for differences in
math score. Unfortunately, due to a technical error, the question asking for math SAT
score did not appear on the questionnaire for the first 16 participants. Given that 4
participants chose to skip this question, of the 64 total participants, math SAT score
was recorded for only 44 participants.

5
Caps and “challenging” included in prompt to emphasize competitive context.

132
Chapter 7: Results
Two regression models with the number of correct math task responses as the
dependent variable were constructed to test the hypotheses and research question. In
the first model, the avatar-assignment conditions, interaction between these two
conditions, and the covariate of SAT score were included as the independent
variables. The composite measures of the three levels of self-presence were added to
the second model. The results are listed in Table 7.1.
Table 7.1. Regression Models on Number of Correct Math Responses

Model 1
Model 2

Cohen's
Cohen's
Beta t d size^ Beta t d size^
Avatar customization
-0.05 -0.38

-0.10 -0.57

Gender consistency
0.03 0.19

0.05 0.33

Interaction (custom *
gender)
-0.26 -1.83 p=.08 0.59 med -0.20 -1.30

Math SAT score
0.46 3.17
**
-1.02 large 0.46 3.10
**
1.03 large
Proto self-presence

-0.29 -1.85
p=.07
-0.62 med
Core self-presence

0.06 0.34

Extended self-
presence

0.08 0.46

df 39 36
R
2

0.24

0.31

Note: † p < .10, *p < .05, **p < .01, ***p < .001
^ >0.8 considered large, 0.2-0.8 medium, <.2
small

These results do not support Hypothesis 1 because there was no difference in
math task performance by gender. However, there was an interaction effect that
approached significance, with participants who used customized, similarly gendered
or generic, oppositely gendered avatars performing worse on the math task than
participants who used generic, similarly gendered or customized, oppositely gendered
avatars.

133
Given that Hypothesis 1 was not supported, it was not possible to test
Hypothesis 2 nor Research Question 1. However, the results from the second
regression model provide an interesting insight into the relationship between self-
presence and performance on the math task. Namely, in a nearly significant
relationship (p = .07), proto self-presence is negatively related to math performance.
Given that the interaction between the avatar-assignment conditions was positively
related to proto self-presence (Chapter 4) and that the relationship between this
interaction and math performance becomes non-significant when proto self-presence
is entered into the model, it follows that the interaction effect on math performance is
entirely mediated by proto self-presence (Baron & Kenny, 1986).

134
Chapter 7: Discussion
The present chapter examines how self-presence influences learning-oriented
game environments. Although the initial intent was to test whether self-presence
plays a role in the stereotype threat-induced relationship between avatar gender and
math performance, this test was not possible because the current study design did not
produce such a relationship. This is especially notable because this experiment
included only female participants and female performance is affected more negatively
than males by stereotype threat (Spencer, Steele, & Quinn, 1999). While this is
unfortunate for the purposes of the study, the positive implication is that the prompt in
which participants were told that their math scores would be compared across avatar
genders was not strong enough to induce stereotype threat. Thus, competition
between avatar genders does not necessarily hinder performance in learning-oriented
games. This is consistent with Lee’s (2009) dissertation, in which stereotype threat
was induced through not only the competitive environment but also the perception
that the participants were in a female minority within the group math-task
environment. In order to replicate and test for the role of self-presence in this
relationship, future research should include both a competitive as well as female-
minority context.
Beyond the lack of stereotype threat induction, the most notable present
finding is that proto self-presence appears to hinder performance on the math task.
This can be interpreted as either a state or a trait effect. Regarding the former, the
experience of proto self-presence may facilitate an embodied style of cognition and
this may hinder subsequent abstract cognition to which the physical body cannot
contribute. Cognitive processes are sometimes integrated with the body and
environment (Clark, 1999), which is why people often use their bodies to manipulate

135
objects in order to facilitate thinking about those objects, e.g., moving Scrabble letter-
tiles around in order to “see” more combinations that yield words (Wilson, 2002). If
someone becomes accustomed to this mode of cognition, then subsequent tasks that
restrict the body’s input may be more difficult for this person. Similarly, someone
who experiences more proto self-presence in the virtual environment may feel
restricted from using a part of their embodied cognitive system when asked to
perform mental math questions. One way to test this explanation would be to allow
participants to use the avatar to perform a cognitive task and then to measure if
performance is influenced by the extent of proto self-presence.
The trait-based explanation of the previous finding is that people who tend to
experience more proto self-presence also tend to perform worse on math tasks in
general. While this may be the most parsimonious explanation, there are two pieces
of evidence that contradict it. First, SAT scores were used as a covariate in the
analysis and so this should have controlled for trait-based differences in math ability.
Second, the interaction effect is clearly causal because participants were randomly
assigned to condition. Thus, the finding that proto self-presence completely mediates
this effect (though only at the p = .07 level) indicates that the only ancillary cause of
proto self-presence that influences math performance is avatar condition.
This suggests an alternative state explanation that is less complex than that
previously proposed. Perhaps proto self-presence is simply positively related to
enjoyment and the more someone enjoys playing a game, the less they are able or
motivated to focus on a learning task afterward. This reasoning is consistent with the
finding that proto self-presence is positively related to zygomatic EMG activity
(smiling) during avatar use (Chapter 5). A Pearson’s correlation indicates that proto
self-presence is significantly (p < .01) related to participants’ ratings (on a 1-5 likert

136
scale) of the extent to which they had fun and enjoyed playing the game (.43 and .36,
respectively). This supports the claim for a negative relationship between enjoyment
and academic focus in a game, which is quite intuitive: people dislike engaging in
boring or arduous tasks immediately after engaging in a fun task. One disconcerting
implication of this explanation is that fun games are inherently limited in educational
ability. But future research on self-presence could help elucidate this relationship by
independently manipulating enjoyment and self-presence in a learning-oriented video
game and then examining learning outcomes.
One additional issue worth noting is that the lack of findings may be due to the
difference in the math task from Lee’s (2009) original task. As noted earlier, instead
of 20 questions given over a span of 10 minutes, the present study assigned 10
questions over a span of five minutes. It is possible that this reduced the variance in
the number answered correctly and thus dampened the stereotype threat effects.
While the other theoretically driven explanations seem more plausible, future research
should still consider this possibility and use a more robust assessment of stereotype
threat effects.
The present chapter contributes the first empirical application of this
dissertation’s framework of self-presence to a phenomenon besides self-presence
itself. While the substantive results of this particular study are small, they support the
claim that self-presence is relevant to outcomes after avatar use and thus that future
research should apply self-presence to a variety of topics. The following chapter
concludes this dissertation by suggesting various approaches for research that can
benefit from and contribute to an understanding of the concept of self-presence.

137
Conclusion: Summary and Future

Conclusion: Abstract
This chapter summarizes the various contributions made to as well as
questions raised about the construct of self-presence by this dissertation. It then
suggests various approaches to research that can benefit from and contribute to an
understanding of the concept of self-presence. These approaches include but are not
limited to improving, reinterpreting, or standardizing our understanding of behavioral
or psychological effects of avatar use as well as effects indirectly related to avatar use.
In conclusion, the concept of self-presence is likely to be valuable within our future
media landscape.

The previous seven chapters present 35 hypotheses and seven research
questions that guide the examination of the three levels of self-presence through
results from five studies with 642 participants. While likely the largest single body of
work to date on the concept of self-presence, this dissertation represents an early
contribution to this concept’s potential ability to improve our understanding and
practice of media use. Thus, after briefly summarizing the dissertation’s main
contributions and questions raised, the current chapter outlines specific examples of
research that can benefit from and contribute to the concept of self-presence. It
concludes with a discussion of the larger implications of this research agenda, the
potential of a full theory of self-presence, and the future media landscape within
which such a theory would be useful.

138
Summary of Contributions: 100+ pages in one paragraph
This dissertation offers a conceptual framework and measurement tool of self-
presence as well as support for the concept’s reliability, validity, and usefulness
across numerous virtual contexts, from collaborative student-oriented virtual
environments to multiplayer online games to a movement-based console game.
Though interrelated with each other, the three levels of self-presence discretely relate
to game genre, avatar customization, avatar personalization, avatar gender
consistency, physiological effects during and after avatar use, and the ability to
perform on a math task after avatar use. These relationships were mostly consistent
with expectations gleaned from the construct’s conceptual framework. The findings
also raise questions about the order of causation, potential mediating variables, state
or trait-based interpretations, generalizability to other populations, detailed temporal
differences between levels of self-presence (e.g., dissipation rate of core self-
presence), and relationships with unanticipated variables (e.g., enjoyment). Future
research that addresses these important questions will contribute to an even greater
understanding of the concept of self-presence.
Future Contributions
Beyond addressing questions targeted at developing a greater understanding of
self-presence itself, the framework and measurement tool offered in this dissertation
are likely to be useful to the growing area of research about the implications of using
avatars. Such research may focus on direct behavioral or psychological effects as
well as effects indirectly related to avatar use. Further, self-presence may help
research build upon or reinterpret previous results in a standardized manner. Specific
examples follow.

139
One recent and popular discovery in the area of avatar use effects, the Proteus
Effect, relates to all three levels of self-presence, although without using the explicit
framework offered here (Yee & Bailenson, 2007). The fundamental point of the
Proteus Effect, that people conform to behavioral expectations imbued by their
avatars’ identity, relates to extended self-presence because it requires an identity-level
connection between the user and the avatar. This effect is reinforced through the
other levels of self-presence. Specifically, researchers induce proto self-presence by
having participants look into a virtual mirror and watch as their avatars’ movements
correspond with their own (Yee & Bailenson, 2007). In one study, researchers
induced core self-presence by asking participants to perform a memory task that made
those participants in older-adult avatars associate their behavior in the virtual world
with typical older-adult emotional responses to such a task (Yee & Bailenson, 2006).
Given this conceptual relationship between the framework of self-presence
and the Proteus Effect, research on the Proteus Effect could use the Self-Presence
Questionnaire (SPQ) to control for individual differences and thereby improve results.
For example, an appropriately timed administration of the SPQ may help explain why
participants who used taller avatars negotiated more aggressively in both virtual world
and subsequent face-to-face interactions during the first but not the second negotiation
session within both environments (Yee, et al., 2009). Specifically, the authors suggest
that a lingering self-perception effect may drive this finding, and thus differences in
the three levels of self-presence, which dissipate at unique rates after avatar use
(Chapter 6), may contribute to this effect. Similarly, responses to the SPQ may help
explain why gender was found to influence the extent to which using a black or white-
cloaked avatar affects group cohesion scores (Pena, et al., 2009). Specifically, this

140
research could confirm the authors’ reasoning that this effect is driven by avatar
gender consistency by including extended self-presence as a covariate.
Outside of the Proteus Effect research paradigm, the concept of self-presence
can also contribute to various other types of research on behavioral effects of avatar
use. For example, avatar characteristics can influence behavior, such as subsequent
exercise, when the avatar owner simply observes but does not control the avatar (Fox
& Bailenson, 2009). Identification with similar-to-self avatars is conceptualized to
drive this finding and the application of a measure of extended self-presence could
test this idea. Further, Fox, Bailenson, & Binney (2009) argued that a similar effect,
with respect to the influence of observing a similar-to-self avatar on eating behaviors,
is related to presence. However, three of the ten questions in their presence measure
actually address self-presence (e.g., “To what extent do you feel you embodied the
avatar?”). Instead of this measure, the SPQ as well as a measure of (virtual
environment-oriented) presence could be used to isolate a more specific cause of this
effect.
The application of self-presence may also contribute to research on the
psychological effects of avatar research. For example, differences between people
who create and use either an ideal-self or actual-self avatar (Jin, 2009) may be
explained by differences between the levels of self-presence with respect to
customization versus personalization (Chapter 2). Specifically, ideal-self avatars may
be more strongly associated with proto and core self-presence, while actual-self
avatars may be more strongly associated with extended self-presence and these
differences may explain differences during or after avatar use.
The concept of self-presence may also help reframe and standardize findings
regarding avatar use. For example, Chandler, Konrath, & Schwarz’s (2009) findings

141
regarding the difference between temporary and chronic accessibility of avatars, as
indicated by the extent of avatar-self overlap, may also be explained by differences in
self-presence. Specifically, the finding that avatar body size is more likely to
influence own body size judgments of people for whom avatar-self overlap is large,
and thus the avatar is chronically (always) accessible, may also be explained by
difference in self-presence. In other words, greater avatar-self overlap or chronic
avatar accessibility may relate to higher degrees of self-presence. While the inclusion
of self-presence in such an examination would not change the interpretation of the
results, it would make them more easily comparable to other findings that explicitly
relate to self-presence.
The inclusion of self-presence may also help with the reinterpretation of
previous findings. For example, the finding that using a racially black avatar
compared to using a racially white avatar increases implicit bias against blacks is not
entirely consistent with expectations (Groom, Bailenson, et al., 2009). The authors
explain that stereotype activation drives this finding, but perhaps this is only true for
people who experience low levels of self-presence. In other words, people who
experience more self-presence may engage in more perspective taking and thus
exhibit less racial bias after using a racially black avatar.
Self-presence may also contribute to the development of research in areas that
have received little attention. For example, as described in Chapter 7, the effects of
avatar use on performance in learning-oriented video games may relate to self-
presence. This research may lead to an understanding of how avatars may hinder
learning processes, such as by inducing stereotype threat (Lee, 2009) or facilitating a
cognitive state that inhibits certain types of thinking (Chapter 7). This research could
lead to a set of suggestions for using avatars to improve performance in learning-

142
oriented games, such as by using increasing avatar customization options to control
the rate at which players develop extended self-presence, thereby motivating
improved performance.
Beyond the specific examples of potential research described here, there are
numerous other types of research that may benefit from the application of self-
presence. Such research may use self-presence examine other facets of avatar use,
from social relationships in online games (Steinkuehler & Williams, 2006) to the
treatment of psychological conditions in virtual environments (Rizzo, et al., 2005).
Self-presence may also be applicable to research on other types of self-representations
besides avatars. For example, the framework of self-presence may help describe how
robots can elicit social connections (Lee, Peng, Jin, & Yan, 2006) or be treated as an
extension of the self (Groom, Takayama, Ochi, & Nass, 2009).
Toward a Theory of Self-Presence
Regardless of the specific application, the benefits of including the concept of
self-presence in future research are likely to be mutual in that they will also contribute
to a greater understanding of self-presence itself. These benefits, regardless of
directionality, are contingent upon the extent to which researchers in this area use the
concept of self-presence as a common tool to facilitate the control and measurement
of the ways and extent to which people feel connected to their avatars. The
fundamental goal of this dissertation has been to test and thereby support the value of
this tool. Thus, while this dissertation’s contribution to the concept and application of
self-presence is relatively nominal, it has laid the early foundations for self-presence
to be a part of larger future contributions to our understanding of avatar use and media
use in general.

143
This raises the question of whether this research trajectory will result in a full
theory of self-presence. Prior to this dissertation, self-presence was a vaguely defined
concept. This dissertation developed it into a framework and then tested the
foundations of this framework. Though the framework withstood these tests so far,
much more exploration is required in order to mold this framework into a full theory.
As described above, future research using the framework of self-presence offered here
will provide further insights into the meaning and value of the framework. A full
theory of self-presence may emerge from such research, synthesizing the results in a
way that is applicable across a wide range of avatar-use scenarios. This theory may
begin with the simple claim that people connect with their self-representations on
three distinct levels of self: body, emotion, and identity. But then the theory may
increase in complexity as it describes differences in the causes for each level of
connection, such as that people experience a greater body-level connection when they
are physically similar to the avatar, but a greater identity-level connection when they
use a customized avatar (Chapter 4). And then the complexity may increase even
further as it describes the effects of each level of connection, such as that body-level
connections to avatars lead to poor cognitive performance (Chapter 7). Because these
examples are based on the findings from the present dissertation, they are most likely
myopic relative to future research on the topic. Perhaps future research will uncover a
simpler continuum through which the concept of self-presence can be described more
parsimoniously.
Regardless of the form that a full theory of self-presence takes, such a theory
will become more valuable as our media landscape increasing facilitates connectivity
to avatars. Currently, avatars are confined mostly to video games and virtual worlds,
but they are likely to increasingly pervade our mediated and even non-mediated life.

144
Further, it is important to remember that every avatar is partly an agent, i.e., some
avatar behavior is controlled by the user and some is automatic (e.g., blinking;
Bailenson & Blascovich, 2004). The balance between controlled and automatic
behavior in avatars is likely to shift as computing resources increasingly facilitate
treating avatars more like agents. In other words, avatars may allow us to take full
control at some times but then act for us as independent entities at other times. An
avatar then becomes both a synchronous and asynchronous representation of the
avatar user. Avatars could be set to autopilot for simple tasks but then users could
inhabit them for more complex tasks. For example, a user may send her avatar on
dates with numerous other avatars. While on autopilot, the avatars can perform basic
screening of the other avatars based on information that is simple enough to process
automatically. Then the user could inhabit the avatar in order to engage in
synchronous communication with the dates who were screened as acceptable. The
potential applications of such avatars become even more interesting given that avatars
may also be manifested physically, i.e., as robots. For example, an individual could
send her robot-avatar
6
to an open house with instructions to compare certain facets of
the home to the user's preferences. After completing this automatic task, the user
could inhabit the robot-avatar and take a closer look at some of the details in the
home. While such a media landscape may seem more like science fiction than reality,
science fiction authors are adept at identifying valuable research questions for media
scholars (Bailenson, Yee, Kim, & Tecarro, 2007) and so such potential futures should
be taken seriously. By developing a greater understanding of how people connect to
their avatars, i.e., self-presence, we may better prepare for a future in which such
avatar use scenarios become a reality.

6
Such robot avatars are already on the market: http://www.anybots.com/

145
Beyond the potential academic and practical applications of a full theory of
self-presence, this stream of research may have important philosophical implications.
Self-presence describes the ways that we connect to our virtual self-representations,
but as such connections become increasingly complex and ubiquitous, the distinction
between the unmediated self and the virtual self loses relevance. Thus, to a
continually evolving extent, to know the virtual self is to “know thyself”. Plato
emphasized this ancient Greek maxim over two thousand years ago, but its meaning is
still relevant to the human experience. Though most of our time is spent looking
outward at the world around us, we must also look inward at ourselves in order to
understand how we relate to this world. Such introspection is inherently difficult.
Our environment bombards us with signals that grab our attention and guide our
decisions, hindering our ability to engage in self-reflection. However, while our
current media landscape proliferates nearly every moment of our daily lives with such
distractions, it also provides us with avatars that have the potential to reflect our
selves right in front of our eyes. In other words, we can use avatars to engage in
introspection while still looking outward, to some extent. Thus, this dissertation’s
contribution to our understanding of avatar use may provide a foundation for avatar-
facilitated self-reflection, thereby improving our experience within our many worlds.

146
References

Alankus, G., Lazar, A., May, M., & Kelleher, C. Towards customizable games for
stroke rehabilitation. CHI 2010: Proceedings of the 28th international
conference on Human factors in computing systems, 2113-2122.
Anderson, C. (2002). Violent video games and aggressive thoughts, feelings and
behaviors. In S. Calvert, A. B. Jordan & R. B. Cocking (Eds.), Children in the
digital age: Influences of electronic media on development. Westport: Praeger.
Armel, K., & Ramachandran, V. (2003). Projecting sensations to external objects:
Evidence from skin conductance response. Proceedings of the Royal Society
B: Biological Sciences, 270, 1499-1506.
Bailenson, J., & Blascovich, J. (2004). Avatars. Encyclopedia of Human-Computer
Interaction. Berkshire Publishing Group, 64–68.
Bailenson, J., Blascovich, J., Beall, A., & Loomis, J. (2001). Equilibrium theory
revisited: Mutual gaze and personal space in virtual environments. Presence,
10(6), 583-598.
Bailenson, J. N., Yee, N., Kim, A., & Tecarro, J. (2007). Sciencepunk: The influence
of informed science fiction on virtual reality research. In M. Grebowicz (Ed.),
The Joy of SF: Essays in Science and Technology Studies (pp. 147-164): Open
Court Publishing.
Baranowski, T., Baranowski, J., Cullen, K. W., Marsh, T., Islam, N., Zakeri, I., et al.
(2003). Squire's Quest!:: Dietary outcome evaluation of a multimedia game.
American Journal of Preventive Medicine, 24(1), 52-61.
Bargh, J. A., McKenna, K. Y. A., & Fitzsimmons, G. M. (2002). Can you see the real
me? Activation and expression of the "true self" on the Internet. Journal of
Social Issues, 58(1), 33-48.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in
social psychological research: Conceptual, strategic, and statistical
considerations. Journal of Personality and Social Psychology, 51(6), 1173.
Bartholomew, L., Gold, R., Parcel, G., Czyzewski, D., Sockrider, M., Fernandez, M.,
et al. (2000). Watch, Discover, Think, and Act: evaluation of computer-
assisted instruction to improve asthma self-management in inner-city children.
Patient Education and Counseling, 39(2-3), 269-280.
Basak, C., Boot, W., Voss, M., & Kramer, A. (2008). Can training in a real-time
strategy video game attenuate cognitive decline in older adults. Psychology
and Aging, 23(4), 765-777.

147

Beale, I. L., Kato, P. M., Marin-Bowling, V. M., Guthrie, N., & Cole, S. W. (2007).
Improvement in cancer-related knowledge following use of a
psychoeducational video game for adolescents and young adults with cancer.
Journal of Adolescent Health, 41(3), 263-270.
Behrend, T. S., & Thompson, L. F. (in press, 2011). Trainer-trainee similarity effects
in web-based training. Computers in Human Behavior.
Bem, S. L. (1981). Gender schema theory: A cognitive account of sex typing.
Psychological review, 88(4), 354.
Ben-Shakhar, G. (1985). Standardization within individuals: A simple method to
neutralize individual differences in skin conductance. Psychophysiology,
22(3), 292-299.
Berardesca, E., & Maibach, H. (1996). Racial differences in skin pathophysiology.
Journal of the American Academy of Dermatology, 34(4), 667-672.
Berti, A., & Frassinetti, F. (2000). When far becomes near: Remapping of space by
tool use. Journal of Cognitive Neuroscience, 12(3), 415-420.
Biocca, F. (1997). The cyborg’s dilemma: Embodiment in virtual environments.
JCMC, 3(2), 12–26.
Bolter, J. (1984). Turing's man: Western culture in the computer age. Chapel Hill,
North Carolina: The University of North Carolina Press.
Botvinick, M., & Cohen, J. (1998). Rubber hands 'feel' touch that eyes see. Nature,
391, 756.
Bushman, B. J., & Huesmann, L. R. (2006). Short-term and long-term effects of
violent media on aggression in children and adults. Archives of Pediatrics &
Adolescent Medicine, 160(4), 348-352.
Castronova, E., Williams, D., Huang, Y., Shen, C., Keegan, B., Ratan, R. A., et al.
(2009). As real as real? Macroeconomic behavior in a large-scale virtual
world. New Media & Society, 11, 685-707.
Chan, E., & Vorderer, P. (2006). Massively multiplayer online games. In P. Vorderer
& J. Bryant (Eds.), Playing computer games - Motives, responses, and
consequences (pp. 77-88). Mahwah, NJ: Lawrence Erlbaum.
Chandler, J., Konrath, S., & Schwarz, N. (2009). Online and on my mind: Temporary
and chronic accessibility moderate the influence of media figures. Media
Psychology, 12(2), 210-226.
Clark, A. (1999). An embodied cognitive science? Trends in Cognitive Sciences, 3(9),
345-351.

148
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale,
New Jersey: Lawrence Erlbaum Associates.
Courtney, C. G., Dawson, M. E., Schell, A. M., Iyer, A., & Parsons, T. D. (2010).
Better than the real thing: Eliciting fear with moving and static computer-
generated stimuli. International Journal of Psychophysiology, 78(2), 107-114.
Damasio, A. R. (1994). Descartes' error: Emotion, reason, and the human brain.
New York, NY: Putnam.
Damasio, A. R. (1999). The feeling of what happens: Body and emotion in the making
of consciousness (1st ed.). New York, NY: Harcourt Brace.
Dawson, M. E., Rissling, A. J., Schell, A. M., & Wilcox, R. (2007). Under what
conditions can human affective conditioning occur without contingency
awareness? Test of the evaluative conditioning paradigm. Emotion, 7(4), 755-
766.
Dawson, M. E., & Schell, A. M. (2002). What does electrodermal activity tell us
about prognosis in the schizophrenia spectrum? Schizophrenia research, 54(1-
2), 87-93.
Dawson, M. E., Schell, A. M., & Filion, D. L. (2000). The electrodermal system. In J.
T. Cacioppo, L. G. Tassinary & G. L. Bernston (Eds.), Handbook of
Psychophysiology (pp. 200-223). Cambridge: Cambridge University Press.
Deutsch, J. E., Borbely, M., Filler, J., Huhn, K., & Guarrera-Bowlby, P. (2008). Use
of a low-cost, commercially available gaming console (Wii) for rehabilitation
of an adolescent with cerebral palsy. Physical Therapy, 88(10), 1196.
Ducheneaut, N., Wen, M. H., Yee, N., & Wadley, G. (2009). Body and mind: a study
of avatar personalization in three virtual worlds. . Proceedings of the 27th
international conference on Human factors in computing systems. 1151-1160.
ACM.
Ehrsson, H., Spence, C., & Passingham, R. (2004). That's my hand! Activity in
premotor cortex reflects feeling of ownership of a limb. Science, 305(5685),
875-877.
Ekman, P. (1971). Universals and cultural differences in facial expressions of
emotion: University of Nebraska Press.
Ekman, P. (1992). Facial expressions of emotion: New findings, new questions.
Psychological Science, 3(1), 34.
Ekman, P. (1993). Facial expression and emotion. American Psychologist, 48, 384-
384.
Erikson, E. (1959). Identity and the life cycle. Selected papers. New York:
International University Press.

149
ESA (2010). Essential facts about the computer and video game industry.
Washington, D.C.: Entertainment Software Association.
Evans, D., & Zarate, O. (1999). Introducing evolutionary psychology. New York, NY:
Totem Books.
Farnè, A., & Làdavas, E. (2000). Dynamic size-change of hand peripersonal space
following tool use. NeuroReport, 11(8), 1645.
Feng, J., Spence, I., & Pratt, J. (2007). Playing an action video game reduces gender
differences in spatial cognition. Psychological Science - Cambridge, 18(10),
850.
Festinger, L. (1957). A theory of cognitive dissonance: Stanford Univ Pr.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with
unobservable variables and measurement error. Journal of marketing research,
18(1), 39-50.
Fox, J., & Bailenson, J. (2009). Virtual self-modeling: The effects of vicarious
reinforcement and identification on exercise behaviors. Media Psychology,
12(1), 1-25.
Fox, J., Bailenson, J., & Binney, J. (2009). Virtual experiences, physical behaviors:
The effect of presence on imitation of an eating avatar. Presence:
Teleoperators and Virtual Environments, 18(4), 294-303.
Gamberini, L., Barresi, G., Majer, A., & Scarpetta, F. (2008). A Game a Day Keeps
the Doctor Away: A short review of Computer Games in Mental Healthcare.
Journal of CyberTherapy and Rehabilitation, 1, 127-146.
Gati, I., & Ben-Shakhar, G. (1990). Novelty and significance in orientation and
habituation: A feature-matching approach. Journal of Experimental
Psychology: General, 119(3), 251.
Gibbs, J., Ellison, N., & Heino, R. (2006). Self-presentation in online personals: The
role of anticipated future interaction, self-disclosure, and perceived success in
Internet dating. Communication Research, 33(2), 152.
Goffman, E. (1959). The presentation of self in everyday life. Garden City, New York:
Doubleday.
Green, C., & Bavelier, D. (2003). Action video game modifies visual selective
attention. Nature, 423, 534-537.
Groom, V., Bailenson, J., & Nass, C. (2009). The influence of racial embodiment on
racial bias in immersive virtual environments. Social Influence, 4(1), 1-18.
Groom, V., Takayama, L., Ochi, P., & Nass, C. (2009). I am my robot: the impact of
robot-building and robot form on operators.

150
Hart, S., & Battiste, V. (1992). Field test of video game trainer. Human Factors and
Ergonomics Society Annual Meeting Proceedings, Training, 36, 1291-1295.
Hayles, N. (1999). How we became posthuman: Virtual bodies in cybernetics,
literature, and informatics: University of Chicago Press.
Higgins, E. (1987). Self-discrepancy: A theory relating self and affect. Psychological
review, 94(3), 319-340.
Hugdahl, K. (1996). Psychophysiology: The mind-body perspective. Cambridge, MA:
Harvard University Press.
Huh, S., & Williams, D. (2010). Dude looks like a lady: Online game gender
swapping. In W. Bainbridge (Ed.), Online worlds: Convergence of the real
and the virtual. New York: Springer.
Hussain, Z., & Griffiths, M. (2008). Gender swapping and socializing in cyberspace:
An exploratory study. CyberPsychology & Behavior, 11(1), 47-53.
Ijsselsteijn, W., de Kort, Y., & Haans, A. (2006). Is this my hand I see before me?
The rubber hand illusion in reality, virtual reality, and mixed reality. Presence:
Teleoperators and Virtual Environments, 15(4), 455-464.
Iriki, A., Tanaka, M., & Iwamura, Y. (1996). Coding of modified body schema during
tool use by macaque postcentral neurones. Neuroreport, 7(14), 2325.
Iriki, A., Tanaka, M., Obayashi, S., & Iwamura, Y. (2001). Self-images in the video
monitor coded by monkey intraparietal neurons. Neuroscience Research,
40(2), 163-173.
Ivory, J. D., & Kalyanaraman, S. (2007). The Effects of Technological Advancement
and Violent Content in Video Games on Players' Feelings of Presence,
Involvement, Physiological Arousal, and Aggression. Journal of
Communication, 57(3), 532.
Jansz, J., & Tanis, M. (2007). Appeal of playing online first person shooter games.
CyberPsychology & Behavior, 10(1), 133-136.
Jin, S. (2010). Does imposing a goal always improve exercise intentions in avatar-
based exergames? The moderating role of interdependent self-construal on
exercise intentions and self-presence. CyberPsychology, Behavior, and Social
Networking, 13(3), 335-339.
Jin, S. A. (2009). Avatars mirroring the actual self versus projecting the ideal self:
The effects of self-priming on interactivity and immersion in an exergame,
Wii Fit. CyberPsychology & Behavior, 12(6), 761-765.
Jin, S. A., & Park, N. (2009). Parasocial interaction with my avatar: Effects of
interdependent self-construal and the mediating role of self-presence in an
avatar-based console game, Wii. Cyberpsychology & Behavior, 12(6), 723-
727.

151
Johnson, L. C., & Landon, M. M. (1965). Eccrine sweat gland activity and racial
differences in resting skin conductance. Psychophysiology, 1(4), 322-329.
Kaiser, L. (1989). Adjusting for basline: Change or percentage change. Statistics in
Medicine, 8, 1183-1190.
Kato, P. M., Cole, S. W., Bradlyn, A. S., & Pollock, B. H. (2008). A video game
improves behavioral outcomes in adolescents and young adults with cancer: A
randomized trial. Pediatrics, 122(2), 305-317.
Kim, J., & Mueller, C. (1978). Factor analysis: Statistical methods and practical
issues. Newbury Park, CA: Sage.
Kim, J. O., & Mueller, C. W. (1988). Introduction to factor analysis: What it is and
how to do it (Vol. 13): Sage.
Lachlan, K., & Krcmar, M. (2011). Experiencing Presence in Video Games: The Role
of Presence Tendencies, Game Experience, Gender, and Time Spent in Play.
Communication Research Reports, 28(1), 27-31.
Lang, A. (1994). What can the heart tell us about thinking? In A. Lang (Ed.),
Measuring psychological responses to media messages. Hillsdale, N.J.:
Lawrence Erlbaum.
Lang, A., Chung, Y., Lee, S., Schwartz, N., & Shin, M. (2005). It's an Arousing, Fast-
Paced Kind of World: The Effects of Age and Sensation Seeking on the
Information Processing of Substance-Abuse PSAs. Media Psychology, 7(4),
421.
Lang, A., Potter, R. F., & Bolls, P. (2009). Where Psychophysiology Meets the
Media. In J. Bryant & M. B. Oliver (Eds.), Media Effects: Advances in Theory
and Research. 3rd Edition. . New York, NY. : Routledge.
Lee, J. R. (2009). A threat on the net: Stereotype threat in avatar-represented online
groups. Unpublished doctoral dissertation. Stanford University.
Lee, K. M. (2004). Presence, explicated. Communication Theory, 14(1), 27-50.
Lee, K. M., & Jung, Y. (2005). Evolutionary nature of virtual experience. Journal of
Cultural and Evolutionary Psychology, 3(2), 159-176.
Lee, K. M., Peng, W., Jin, S. A., & Yan, C. (2006). Can robots manifest personality?:
An empirical test of personality recognition, social responses, and social
presence in human–robot interaction. Journal of Communication, 56(4), 754-
772.
Li, R., Polat, U., Makous, W., & Bavelier, D. (2009). Enhancing the contrast
sensitivity function through action video game training. Nature Neuroscience,
12(5), 549-551.

152
Liao, D., Barnes, R. W., Chambless, L. E., & Simpson, R. J. (1995). Age, race, and
sex differences in autonomic cardiac function measured by spectral analysis of
heart rate variability--The ARIC study* 1. The American journal of
cardiology, 76(12), 906-912.
Lim, S., & Reeves, B. (2009). Being in the game: Effects of avatar choice and point of
view on psychophysiological responses during play. Media Psychology, 12(4),
348-370.
Lombard, M., & Ditton, T. (1997). At the heart of it all: The concept of presence.
Journal of Computer-mediated Communication, 3(2), 20.
Lombard, M., Reich, R., Grabe, M., Bracken, C., & Ditton, T. (2000). Presence and
television. Human Communication Research, 26(1), 75-98.
Lovibond, P. F. (1992). Tonic and phasic electrodermal measures of human aversive
conditioning with long duration stimuli. Psychophysiology, 29(6), 621-632.
Lykken, D. T., Rose, R., Luther, B., & Maley, M. (1966). Correcting
Psychophysiological measures for individual differences in range.
Psychological Bulletin, 66(6), 481-484.
MacCallum-Stewart, E. (2008). Real boys carry girly epics: Normalising gender
bending in online games. Eludamos. Journal for Computer Game Culture,
2(1), 27-40.
Maravita, A., & Iriki, A. (2004). Tools for the body (schema). Trends in Cognitive
Sciences, 8(2), 79-86.
Markovsky, B. (1988). Injustice and arousal. Social Justice Research, 2(3), 223-233.
Markus, H., & Nurius, P. (1987). Possible selves: The interface between motivation
and the self-concept. In K. Yardley & T. Honess (Eds.), Self and identity:
Psychosocial perspectives (pp. 157-172). Oxford, England: John Wiley &
Sons.
Matte, J. A. (1996). Forensic psychophysiology using the polygraph: scientific truth
verification, lie detection: JAM Pubns.
Nacke, L. E., Nacke, A., & Lindley, C. A. (2009). Brain training for silver gamers:
effects of age and game form on effectiveness, efficiency, self-assessment, and
gameplay experience. CyberPsychology & Behavior, 12(5), 493-499.
Nagel, T. (1974). What is it like to be a bat? The Philosophical Review, 435-450.
Nah, F. F. H., DeWester, D., & Eschenbrenner, B. (2010). Understanding Gender
Differences in Media Perceptions: A Comparison of 2D versus 3D Media.
SIGHCI 2010 Proceedings, Paper 10.
Ng, B., & Wiemer-Hastings, P. (2005). Addiction to the Internet and online gaming.
CyberPsychology & Behavior, 8(2), 110-114.

153
Pena, J., Hancock, J., & Merola, N. (2009). The priming effects of avatars in virtual
settings. Communication Research, 36(6), 1-19.
Peng, W., Lee, M., & Heeter, C. (2010). The effects of a serious game on role taking
and willingness to help. Journal of Communication, 60(4), 723-742.
Persky, S., & Blascovich, J. (2007). Immersive virtual environments versus traditional
platforms: Effects of violent and nonviolent video game play. Media
Psychology, 10(1), 135-156.
Potter, R. F., & Choi, J. (2006). The Effects of Auditory Structural Complexity on
Attitudes, Attention, Arousal, and Memory. Media Psychology, 8(4), 395.
Ratan, R., & Hasler, B. (2010). Exploring self-presence in collaborative virtual teams.
PsychNology Journal, 8(1), 11-31.
Ratan, R. A. (2010). Self-presence, explicated. Presented at the 60th Annual
Conference of the International Communication Association, Singapore.
Ratan, R. A. (2011). Wii Experiment with Self-Presence: When Mii is Me. To be
presented at the 61st Annual Conference of the International Communication
Association. Boston, MA.
Ratan, R. A., & Hasler, B. (2009). Self-presence standardized: Introducing the Self-
Presence Questionnaire (SPQ). Proceedings of the 12
th
Annual International
Workshop on Presence.
Ratan, R. A., & Ritterfeld, U. (2009). Towards a psychological classification of
serious games. In U. Ritterfeld, M. Cody & P. Vorderer (Eds.), Serious
Games: Mechanisms and Effects (pp. 37-54). New York, NY: Routledge.
Ratan, R. A., Santa-Cruz, M., & Vorderer, P. (2007). Multitasking, presence & self-
presence on the wii. Proceedings of the 10
th
Annual International Workshop
on Presence.
Ravaja, N. (2004). Contributions of Psychophysiology to Media Research: Review
and Recommendations. Media Psychology.
Ravaja, N., Saari, T., Kallinen, K., & Laarni, J. (2006). The Role of Mood in the
Processing of Media Messages From a Small Screen: Effects on Subjective
and Physiological Responses. Media Psychology, 8(3), 239.
Ravaja, N., Saari, T., Salminen, M., Laarni, J., & Kallinen, K. (2006). Phasic
emotional reactions to video game events: a psychophysiological
investigation. Media Psychology, 8(4), 343-367.
Ringle, C. M., Wende, S., & Alexander, W. (2005). SmartPLS 2.0 (beta). Hamburg,
Germany: http://www.smartpls.de.

154
Riva, G., Waterworth, J., & Waterworth, E. (2004). The layers of presence: A bio-
cultural approach to understanding presence in natural and mediated
environments. CyberPsychology & Behavior, 7(4), 402-416.
Rizzo, A., Pair, J., Mcnerney, P., Eastlund, E., Manson, B., Gratch, J., et al. (2005).
Development of a VR therapy application for Iraq war military personnel with
PTSD. Studies in health technology and informatics, 111, 407-413.
Robbins, S., & Bell, M. (2008). Second Life for dummies. Hoboken, New Jersey:
Wiley Publishing, Inc.
Roberts, L., & Parks, M. (1999). The social geography of gender-switching in virtual
environments on the Internet. Information, Communication & Society, 2(4),
521-540.
Rosnow, R. L., & Rosenthal, R. (1996). Computing contrasts, effect sizes, and
counternulls on other people's published data: General procedures for research
consumers. Psychological Methods, 1(4), 331.
Rosser Jr, J. C., Lynch, P. J., Cuddihy, L., Gentile, D. A., Klonsky, J., & Merrell, R.
(2007). The impact of video games on training surgeons in the 21st century.
Archives of Surgery, 142(2), 181.
Salguero, R. A. T., & Moran, R. M. B. (2002). Measuring problem video game
playing in adolescents. Addiction, 97(12), 1601-1606.
Sherman, S. (1997). Perils of the princess: Gender and genre in video games. Western
Folklore, 56(3), 243-258.
Sherry, J. L. (2004). Flow and media enjoyment. Communication Theory, 14(4), 328-
347.
Slater, M., Perez-Marcos, D., Ehrsson, H., & Sanchez-Vives, M. (2008). Towards a
digital body: The virtual arm illusion. Frontiers in Human Neuroscience, 2(6).
Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women's
math performance. Journal of Experimental Social Psychology, 35, 4-28.
Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity
and performance. American Psychologist, 52(6), 613.
Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test
performance of African Americans. Journal of Personality and Social
Psychology, 69(5), 797.
Steinkuehler, C., & Williams, D. (2006). Where everybody knows your (screen)
name: Online games as "third places". Journal of Computer-Mediated
Communication, 11(4).
Steuer, J. (1992). Defining virtual reality: Factors determining telepresence. Journal
of Communication, 42(4), 73–93.

155
Tamborini, R., & Skalski, P. (2006). The role of presence in the experience of
electronic games. In P. Vorderer & J. Bryant (Eds.), Playing video games:
Motives, responses, and consequences (pp. 225–240). Mahwah, NJ: Lawrence
Erlbaum Associates.
Tsakiris, M., & Haggard, P. (2005). The rubber hand illusion revisited: Visuotactile
integration and self-attribution. Journal of Experimental Psychology-Human
Perception and Performance, 31(1), 80-91.
Turkle, S. (1995). Life on the screen: Identity in the age of the Internet. New York:
Touchstone.
Vrana, S. R., & Rollock, D. (2002). The role of ethnicity, gender, emotional content,
and contextual differences in physiological, expressive, and self-reported
emotional responses to imagery. Culture and emotion, 16(1), 165-192.
Warburton, D. E. R., Bredin, S. S. D., Horita, L. T. L., Zbogar, D., Scott, J. M., Esch,
B. T. A., et al. (2007). The health benefits of interactive video game exercise.
Applied Physiology, Nutrition, and Metabolism, 32(4), 655-663.
Waterworth, J., & Waterworth, E. (2003). The core of presence: Presence as
perceptual illusion. Presence Connect, 3(3).
Weber, R., Ritterfeld, U., & Mathiak, K. (2006). Does playing violent video games
induce aggression? Empirical evidence of a functional magnetic resonance
imaging study. Media Psychology, 8(1), 39-60.
Williams, D., Yee, N., & Caplan, S. (2008). Who plays, how much, and why? A
behavioral player census of a virtual world. Journal of Computer Mediated
Communication, 13(4), 993-1018.
Wilson, M. (2002). Six views of embodied cognition. Psychonomic Bulletin and
Review, 9(4), 625-636.
Wise, K., & Reeves, B. (2007). The Effect of User Control on the Cognitive and
Emotional Processing of Pictures. Media Psychology, 9(3), 549-566.
Woodcock, B. S. (2008, April 9). MMOG Subscriptions Market Share - April 2008.
Retrieved October 28, 2009, from http://www.mmogchart.com/Chart7.html
Yee, N. (2006). The demographics, motivations and derived experiences of users of
massively-multiuser online graphical environments. PRESENCE:
Teleoperators and Virtual Environments, 15, 309-329.
Yee, N. (2008). Maps of digital desires: Exploring the topography of gender and play
in online games. In Y. Kafai, C. Heeter, J. Denner & J. Sun (Eds.), Beyond
Barbie and Mortal Kombat: New perspectives on gender and gaming (pp. 83-
96). Cambridge, MA: The MIT Press.

156
Yee, N., & Bailenson, J. (2006). Walk a mile in digital shoes: The impact of
embodied perspective-taking on the reduction of negative stereotyping in
immersive virtual environments. Proceedings of PRESENCE, 24–26.
Yee, N., & Bailenson, J. (2007). The Proteus effect: Self transformations in virtual
reality. Human Communication Research, 33(3), 271-290.
Yee, N., Bailenson, J. N., & Ducheneaut, N. (2009). The Proteus effect: Implications
of transformed digital self-representation on online and offline behavior.
Communication Research, 36(2), 285-312.
Zhao, S., Grasmuck, S., & Martin, J. (2008). Identity construction on Facebook:
Digital empowerment in anchored relationships. Computers in Human
Behavior, 24(5), 1816-1836.

157
Appendix A: Chapter 2 Appendix

Self-Presence Questionnaire (Study 1 – Pilot)

Proto Self-Presence
• When using your avatar, do you feel physically close to the objects and other
avatars in the virtual environment?
• [not at all/somewhat/moderately/very much/a great deal]
• When playing the game/using the virtual environment, how much do you feel
like your avatar is an extension of your body within the game/virtual
environment?
• When something happens to your avatar’s body, to what extent does it feel
like it is happening to any part of your body?
• When using your avatar, to what extent do you feel like your arm is elongated
into the game/virtual environment through your avatar?
• When using your avatar, to what extent do you feel like you can reach into the
game/virtual environment through your avatar?
• When playing the game/using the virtual environment, to what extent do you
feel like your hand is inside of the game/virtual environment?
• When playing the game/using the virtual environment, how much do you feel
your avatar is a part of your body?

Core Self-Presence
• When happy events happen to your avatar, to what extent do you feel happy?
o [not at all/somewhat/moderately/very/extremely]
• When surprising events happen to your avatar, to what extent do you feel
surprised?
• When sad events happen to your avatar, to what extent do you feel sad?
• When upsetting events happen to your avatar, to what extnt do you feel angry?
• When arousing events happen to your avatar, to what extent do you feel
aroused?

Extended Self-Presence
• How much effort did you put into making your avatar’s sex clear to others?
• How much effort did you put into making your avatar’s race clear to others?
• How important is it for your profile to portray a specific identity for your
avatar?
• How much do you care about the age of your avatar?
• How much do you care about the race of your avatar?
• To what extent has the experience of using your avatar helped you learn more
about your own identity?
• To what extent did you customize your avatar to make it look the way it does?
• To what extent does your avatar’s appearance represent some aspect of your
identity?
• How much do you care about the sex of your avatar?
• How much effort did you put into making your avatar’s age clear to others?
• How much do you care about how your avatar looks?

158

Self-Presence Questionnaire (Study 2)

Proto Self-Presence
• When playing the game/using the virtual environment, how much do you feel
your avatar is a part of your body?
o [not at all/somewhat/moderately/very much/absolutely]
• When using your avatar, to what extent do you feel like your arm is elongated
into the game/virtual environment through your avatar?
• When using your avatar, to what extent do you feel like you can reach into the
game/virtual environment through your avatar?
• When playing the game/using the virtual environment, how much do you feel
like your avatar is an extension of your body within the game/virtual
environment?

Core Self-Presence
• When upsetting events happen to your avatar, to what extent do you feel
angry?
o [not at all/somewhat/moderately/very/extremely]
• When disgusting events happen to your avatar, to what extent do you feel
disgusted?
• When arousing events happen to your avatar, to what extent do you feel
aroused?
• When surprising events happen to your avatar, to what extent do you feel
surprised?
• When scary events happen to your avatar, to what extent do you feel afraid?

Extended Self-Presence
• To what extent is your avatar’s gender related to some aspect of your personal
identity?
o [not at all/somewhat/moderately/very much/absolutely]
• To what extent is your avatar’s sex related to some aspect of your personal
identity?
• To what extent is your avatar’s race related to some aspect of your personal
identity?
• To what extent is your avatar’s clothing related to some aspect of your
personal identity?
• To what extent does your avatar’s profile info represent some aspect of your
personal identity?
• To what extent does your avatar’s name represent some aspect of your
personal identity?
• To what extent is your avatar’s appearance related to some aspect of your
personal identity?

159
Self-Presence Questionnaire (Study 3)

Proto Self-Presence
• When playing the game/using the virtual environment, how much do you feel
your avatar is a part of your body?
o [not at all/somewhat/moderately/very much/absolutely]
• When using your avatar, to what extent do you feel like your arm is elongated
into the game/virtual environment through your avatar?
• When using your avatar, to what extent do you feel like you can reach into the
game/virtual environment through your avatar?
• When playing the game/using the virtual environment, how much do you feel
like your avatar is an extension of your body within the game/virtual
environment?
• When playing the game/using the virtual environment, to what extent do you
feel like your hand is inside of the game/virtual environment?

Core Self-Presence – Included in factor analysis
• When upsetting events happen to your avatar, to what extent do you feel
angry?
o [not at all/somewhat/moderately/very/extremely]
• When disgusting events happen to your avatar, to what extent do you feel
disgusted?
• When arousing events happen to your avatar, to what extent do you feel
aroused?
• When surprising events happen to your avatar, to what extent do you feel
surprised?
• When scary events happen to your avatar, to what extent do you feel afraid?
• When happy events happen to your avatar, to what extent do you feel happy?
• When sad events happen to your avatar, to what extent do you feel sad?

Extended Self-Presence
• To what extent is your avatar’s gender related to some aspect of your personal
identity?
o [not at all/somewhat/moderately/very much/absolutely]
• To what extent is your avatar’s race related to some aspect of your personal
identity?
• To what extent is your avatar’s clothing related to some aspect of your
personal identity?
• To what extent does your avatar’s profile info represent some aspect of your
personal identity?
• To what extent does your avatar’s name represent some aspect of your
personal identity?
• To what extent is your avatar’s appearance related to some aspect of your
personal identity?
• To what extent is your avatar’s sex related to some aspect of your personal
identity? (removed from second iteration of analysis)

160
Appendix B: Chapter 3 Appendix

Self-Presence Questionnaire (only items included in this study)
Proto Self-Presence
• When playing the game, how much do you feel like your avatar is an
extension of your body within the game?
o [not at all/somewhat/moderately/very much/absolutely]
• When playing the game, how much do you feel your avatar is a part of your
body?
• When using your avatar, to what extent do you feel like you can reach into the
game through your avatar?
• When using your avatar, to what extent do you feel like your arm is elongated
(changed to “stretched” in Study 2) into the game through your avatar?
• When playing the game, to what extent do you feel like your hand is inside of
the game?

Core Self-Presence
• When happy events happen to your avatar, to what extent do you feel happy?
o [not at all/somewhat/moderately/very/extremely]
• When sad events happen to your avatar, to what extent do you feel sad?
• When surprising events happen to your avatar, to what extent do you feel
surprised?
• When scary events happen to your avatar, to what extent do you feel afraid?
• When upsetting events happen to your avatar, to what extent do you feel
angry?
• When disgusting events happen to your avatar, to what extent do you feel
disgusted?

Extended Self-Presence
• To what extent is your avatar’s appearance related to some aspect of your
personal identity?
o [not at all/somewhat/moderately/very much/absolutely]
• To what extent does your avatar’s name represent some aspect of your
personal identity?
• To what extent is your avatar’s race related to some aspect of your personal
identity?
• To what extent is your avatar’s clothing related to some aspect of your
personal identity?
• To what extent does your avatar’s profile info represent … identity?
• To what extent do you identify with your avatar?

161
Game Coding (number of respondents in parentheses)
Massively Multiplayer Online Games:
Aion (6), Diablo II (1), Dungeons and Dragons Online (1), Guild Wars (1), Ragnarok
Online (2), World of Warcraft (268)

First Person Shooter Games:
Call of Duty* (5), Counter Strike (9), Fallout 3 (1), Half-Life: Natural Selection (1),
Gears of War*(1), Grand Theft Auto* (1), Halo* (3), Left 4 Dead* (4), Team Fortress
2 (1), Uncharted 2 (1), Urban Terror (1)

Other (non included in analysis):
Active Worlds (1), FIFA* (1), Geometry Wars (1), Poker (1), Pro Evolution Soccer
(1), Second Life (23), Starcraft (2)

*Specified as “any in series”

162
Appendix C: Chapter 4 Appendix

Self-Presence Questionnaire (only items included in this study)
Proto Self-Presence
6. When playing the game, how much do you feel like your avatar is an
extension of your body within the game?
a. [not at all/somewhat/moderately/very much/absolutely]
7. When playing the game, how much do you feel your avatar is a part of your
body?
8. When using your avatar, to what extent do you feel like you can reach into the
game through your avatar?
9. When using your avatar, to what extent do you feel like your arm is elongated
(changed to “stretched” in Study 2) into the game through your avatar?
10. When playing the game, to what extent do you feel like your hand is inside of
the game?

Core Self-Presence
2. When sad events happen to your avatar, to what extent do you feel sad?
1. [not at all/somewhat/moderately/very/extremely]
3. When scary events happen to your avatar, to what extent do you feel afraid?
4. When arousing events happen to your avatar, to what extent do you feel
aroused?

Extended Self-Presence
2. To what extent is your avatar’s appearance related to some aspect of your
personal identity?
1. [not at all/somewhat/moderately/very much/absolutely]
3. To what extent does your avatar’s name represent some aspect of your
personal identity?
4. To what extent is your avatar’s race related to some aspect of your personal
identity?
5. To what extent is your avatar’s clothing related to some aspect of your
personal identity?
6. To what extent did your avatar’s skin color represent some aspect of your
personal identity?

163
Appendix D: Chapter 7 Appendix

Math Questions

526 – 258
475 + 347
135 + 498
622 – 245
262 + 178

4378 – 2579
71 x 26
6859 + 2354
3413 – 1677
148 + 595

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Using virtual humans during clinical interviews: examining pathways to self-disclosure

PDF

Virtually human? Negotiation of (non)humanness and agency in the sociotechnical assemblage of virtual influencers

PDF

Penrose Station: an exploration of presence, immersion, player identity, and the unsilent protagonist

PDF

The role of social control in promoting healthy eating behavior among Chinese immigrants: an ecological approach

PDF

Intelligent near-optimal resource allocation and sharing for self-reconfigurable robotic and other networks

PDF

Self-perceptions of Aging in the Context of Neighborhood and Their Interplay in Late-life Cognitive Health

Asset Metadata

Creator Ratan, Rabindra Ayyan (author)

Core Title Self-presence: body, emotion, and identity extension into the virtual self

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Publication Date 07/24/2011

Defense Date 05/20/2011

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag avatar,Collaborative Virtual Environment,computer-mediated communication,console game,gender,human-computer interaction,OAI-PMH Harvest,online game,presence,Proteus Effect,psychophysiology,self-presence,stereotype threat,video game,virtual reality,Wii

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Williams, Dmitri (committee chair), Dawson, Michael E. (committee member), Lee, Kwan Min (committee member)

Creator Email raratan@gmail.com,rratan@stanfordalumni.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c127-642685

Unique identifier UC1382582

Identifier usctheses-c127-642685 (legacy record id)

Legacy Identifier etd-RatanRabin-161-0.pdf

Dmrecord 642685

Document Type Dissertation

Rights Ratan, Rabindra Ayyan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu