THE PSYCHOPHYSIOLOGICAL - QUT · The psychophysiological method presents one approach for evaluating this player experience. A variety of psychophysiological investigations of the

THE PSYCHOPHYSIOLOGICAL

EVALUATION OF THE PLAYER

EXPERIENCE

Madison Klarkowski

B. Games & Interactive Entertainment, Honours (IT).

Written under the supervision of

Assoc. Prof. Daniel Johnson

&

Assoc. Prof. Peta Wyeth

Assoc. Prof. Simon Smith

Submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

School of Electrical Engineering and Computer Science

Science and Engineering Faculty

Queensland University of Technology

2017

i

The Psychophysiological Evaluation of the Optimal Player Experience

Keywords

Challenge; Challenge‒skill balance; Electrocardiography; Electrodermal activity;

Electroencephalography; Electromyography; Enjoyment; Flow; Physiology; Player experience;

Psychophysiology; Self-determination theory; Video games

ii


Abstract

As video games emerge as a leading form of entertainment, so too does the need for a

comprehensive understanding of the player experience. Player experience research thus expands

upon this understanding through the lens of psychological constructs such as flow, presence,

challenge, competence and self-determination theory.

The psychophysiological method presents one approach for evaluating this player

experience. A variety of psychophysiological investigations of the player experience have been

undertaken, and have contributed novel results that expand upon the understanding of

physiological response to video game play. However, these assessments often feature small

sample sizes (occasionally only comprising participants of a single gender), or are restricted to

employing one or two psychophysiological measures. The value of a study with a large sample

size and multiple psychophysiological and subjective measures was thus identified, and

undertaken for this program of research.

For this study, pilot testing was undertaken to confirm the suitability of the chosen

psychophysiological measures, refine the game artefacts and identify the most appropriate

subjective measures to use. The full study was conducted with 90 participants playing three game

conditions that were manipulated in terms of challenge‒skill balance. These conditions featured

one optimal player experience, ‘Balance’ (in which the challenges of the game matched the skills

of the player’), and two sub-optimal player experiences, ‘Overload’ and ‘Boredom’ (in which the

challenges of the game outstripped, or were outstripped by, the skills of the player). The full study

featured the use of both subjective (Player Experience of Needs Satisfaction scale [PENS], flow

and Intrinsic Motivation Inventory [IMI]) and psychophysiological (electrodermal activity [EDA],

iii


electromyography [EMG], electrocardiography [ECG] and electroencephalography [EEG])

measures.

Psychophysiological assessment revealed increased positively valenced emotional

expressivity associated with increased challenge of the condition; greater EDA was found in the

Overload condition than in the Boredom condition; and decreased high-frequency (HF) peak

components of heart rate variability (HRV) were found in the Overload condition compared with

either Boredom or Balance. Results also revealed increased heart rate (HR) in the Boredom

condition. Greater EEG alpha, beta and theta activity was also found in Balance and Overload

conditions.

These results suggest increased positive valence with challenge, and greater presence of

arousal in the Balance (optimal) and Overload conditions; however, increased HR in the Boredom

condition indicates some complexities in interpreting psychophysiological data or assessing sub-

optimal player experiences. Results for cognitive activity suggest greater alertness, creativity,

attentional focus, problem-solving and restfulness in the Balance and Overload conditions, as

assessed by electroencephalographic alpha, beta and theta frequency bands. Predictive

relationships between physiological responses and specific subjective responses were not found,

suggesting that psychophysiological evaluation may be limited in predicting individual

components of the player experience.

Overall, this study identifies psychophysiological evaluation as an insightful and

distinctive approach for assessing the player experience. It proposes recommendations for

employing this approach alongside subjective analysis. Despite this, limitations exist for using

psychophysiological evaluation in terms of its temporal, financial and methodological costs;

however, these limitations may be minimised through certain approaches.

iv


List of Publications

Klarkowski, M., Johnson, D., Wyeth, P., McEwan, M., Phillips, C., & Smith, S. (2016).

Operationalising and Evaluating Sub-Optimal and Optimal Play Experiences through

Challenge‒Skill Manipulation. Proceedings of the 2016 CHI Conference on Human

Factors in Computing Systems (CHI ’16), 5583‒5594, ACM, Santa Clara, CA. doi:

10.1145/2858036.2858563

Klarkowski, M., Johnson, D., Wyeth, P., Phillips, C., & Smith, S. (2016). Psychophysiology of

Challenge in Play: EDA and Self-Reported Arousal. Proceedings of the 2016 CHI

Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA ’16),

1930‒1936, ACM, Santa Clara, CA. doi: 10.1145/2851581.2892485

Klarkowski, M., Johnson, D., Wyeth, P., Smith, S., & Phillips, C. (2015). Operationalising and

Measuring Flow in Video Games. Proceedings of the Annual Meeting of the

Australian Special Interest Group for Computer Human Interaction (OzCHI ’15), 114

118, ACM, Melbourne, Australia. doi: 10.1145/2838739.2838826

Other publications include:

Vella, K., Cheng, V. W. S., Johnson, D., Mitchell, J., Davenport, T., Klarkowski, M., &

Phillips, C. (2017). Pokémon GO and Social Connectedness. Manuscript submitted for

publication.

Vella, K., Klarkowski, M., Johnson, D., Hides, L., & Wyeth, P. (2016). The social context of

video game play: Challenges and strategies. Proceedings of the Designing Interactive

Systems Conference (DIS ’16), 761‒772, ACM, Brisbane, Australia. doi:

10.1145/2901790.2901823

Phillips, C., Johnson, D., Wyeth, P., Hides, L., & Klarkowski, M. (2015). Redefining

Videogame Reward Types. Proceedings of the Annual Meeting of the Australian Special

Interest Group for Human Interaction (OzCHI ’15), 83‒91, ACM, Melbourne,

Australia. doi: 10.1145/2838739.2838782

Johnson, D., Wyeth, P., Clark, M., & Watling, C. (2015). Cooperative Game Play with Avatars

and Agents: Differences in Brain Activity and the Experience of Play. Proceedings of

the 33rd Annual ACM Conference on Human Factors in Computing Systems (CHI ’15),

3721‒3730, ACM, Seoul, Republic of Korea. doi: 10.1145/2702123.2702468

https://doi.org/10.1145/2858036.2858563

https://doi.org/10.1145/2851581.2892485

v


List of Figures

Figure 1. The flow channel .......................................................................................................... 11

Figure 2. Divisions of the nervous system. .................................................................................. 23

Figure 3. Russell's Two-Dimensional Model of Emotion ............................................................ 28

Figure 4. The relation of valence and arousal. ............................................................................. 29

Figure 5. Relationships between psychological and physiological domains ............................... 35

Figure 6. Illustrations of various phasically occurring physiological measures .......................... 38

Figure 7. Optimal placement of EDA electrodes on palmar sites ................................................ 40

Figure 8. Schematic representation of facial musculature ........................................................... 41

Figure 9. Suggested facial EMG electrode placement ................................................................. 42

Figure 10. A typical ECG trace and the associated physiological events .................................... 44

Figure 11. EEG traces during various mental states .................................................................... 47

Figure 12. International 10‒20 System ........................................................................................ 51

Figure 13. EDA during play and interviews ................................................................................ 55

Figure 14. Research stages. .......................................................................................................... 78

Figure 15. Screenshot of Left 4 Dead 2. ...................................................................................... 94

Figure 16. Left 4 Dead 2: Tank and Witch boss enemies. ........................................................... 97

Figure 17. Screenshot of Boredom condition (second iteration). ................................................ 99

Figure 18. Screenshot of Balance condition. ............................................................................. 100

Figure 19. Screenshot of Overload condition. .......................................................................... 101

Figure 20. Screenshot of tutorial. ............................................................................................... 103

Figure 21. Boredom condition (second iteration), feat. no combat............................................ 106

Figure 22. Screenshot of Sequencer main menu. ....................................................................... 110

Figure 23. Experimental laboratory ........................................................................................... 118

Figure 24. Study 1: Experiment procedure. ............................................................................... 119

Figure 25. Study 1: Flow State Scale total flow results. ............................................................ 122

Figure 26. Study 1: Flow State Scale subscale results ............................................................... 122

Figure 27. Experimental procedure. ........................................................................................... 135

Figure 28. EMG placement ........................................................................................................ 138

Figure 29. ECG placement. ........................................................................................................ 139

Figure 30. TestBench contact quality display. ........................................................................... 141

Figure 31. EEG placement. ........................................................................................................ 141

Figure 32. EDA placement. ....................................................................................................... 143

Figure 33. EMG data comparison .............................................................................................. 143

Figure 34. Study 2 Short Flow State Scale results. .................................................................... 152

Figure 35. IMI Interest/Enjoyment results. ................................................................................ 152

Figure 36. PENS competence results. ........................................................................................ 153

Figure 37. PENS presence results. ............................................................................................. 153

Figure 38. PENS autonomy results. ........................................................................................... 153

Figure 39. EDA results .............................................................................................................. 157

Figure 40. HRV HF Peaks results .............................................................................................. 157

vi


Figure 41. HR results.................................................................................................................. 157

Figure 42. EEG frequency band results (reversed). ................................................................... 159

Figure 43. EMG OO results (reversed). ..................................................................................... 160

vii


List of Tables

Table 1. EEG frequency bands and associated states. ................................................................. 49

Table 2. Overview of research discussed in Chapter 2 ................................................................ 73

Table 3. Game condition differences (second iteration) ............................................................ 101

Table 4. Game condition differences (third iteration) ................................................................ 107

Table 5. Summary of main effect on subjective response ......................................................... 151

Table 6. Summary of main effect on EDA, HR, and HF Peaks ................................................. 156

Table 7. Correlations between subjective and psychophysiological measures .......................... 163

Table 8. Variances for all variables within conditions ............................................................... 164

Table 9. Effect sizes for psychophysiological measures ............................................................ 193

Table 10. Effect sizes for all subjective measures ..................................................................... 194

Table 11. Overview of significant results for subjective measures ........................................... 197

Table 12. Overview of significant results for psychophysiological measures ........................... 201

Table 13. Overview of temporal costs for psychophysiological measures ................................ 206

Table 14. Overview of utility of psychophysiological measurements ....................................... 208

viii


List of Abbreviations & Terms

AI Artificial Intelligence

ANS Autonomic nervous system

Arousal The intensity with which emotions are experienced

Baseline The natural or resting physiological state

BP Blood pressure

BPM (Heart) beats per minute

Challenge‒skill balance In which the challenge of the task does not outstrip the

skill of the player

CS

Corrugator supercilii, muscle located on the brow;

generally, a measure of negatively valenced emotional

expressivity

DDA Dynamic Difficulty Adjustment

Directional fractionation In which one measure of arousal may decrease even as the

other increases

ECG Electrocardiography: measurement of electrical changes

that occur during the heart’s contractions

EDA Electrodermal activity: measurement of electrical activity

of the skin

ix


EEG Electroencephalography: measurement of cortical activity

EMG Electromyography: measure of electrical activity of

muscles

First-person shooter Shooter game in which the camera perspective represents

the player’s point of view

Flow Total absorption in an activity—the ‘optimal experience’

FSS-2 & S FSS-2 Long Flow State Scale-2 & Short Flow State Scale-2

GEQ Game Experience Questionnaire

Habituation Acclimatisation to a stimuli, such that physiological

response diminishes

HCI Human‒Computer Interaction

HF High frequency; in HRV, this occurs between 0.15 and

0.4Hz

HR Heart rate: generally a measurement of beats per minute

HRV Heart rate variability: measurement of the variation of

beat-to-beat intervals

IBI Interbeat interval: period of time that occurs between each

heartbeat

IMI Intrinsic Motivation Inventory

x


LF Low frequency; in HRV, this occurs between 0.04 and

0.15Hz

NPCs Non-player characters

OO Orbicularis oculi: muscle located near the eye; generally, a

measure of positively valenced emotional expressivity

Orienting response Physiological responses that occur when one is orienting

to a new stimulus or environment

PANAS Positive and Negative Affect Schedule

PENS Player Experience of Needs Satisfaction scale

Phasic Analysis of discrete physiological responses to specific

events or stimuli

Player‒character The character in the video game that the player is

controlling

PNS Parasympathetic nervous system

Presence The experience of ‘feeling there’, being transported to the

mediated world

Psychophysiology The study of relationships between psychological states

and physiological responses

REM sleep Rapid eye movement sleep

SAM Self-Assessment Manikin

xi


SNS Sympathetic nervous system

SPQ Spatial Presence Questionnaire

Startle response

A pattern of physiological responses (e.g., reflexive

blinking) that may occur in reaction to startling stimuli

(e.g., a thunder crack)

Stimulus-response specificity A pattern of physiological responses that occur in reaction

to specific stimuli

Tonic Ongoing analysis of physiological response to a stimuli;

an 'average' of the experience

Valence Positivity or negativity of emotional experience

Video game Digital game played on a computer, tablet, mobile phone

or dedicated gaming console (e.g., PlayStation 4)

VLF Very low frequency; in HRV, this occurs between 0 and

0.04Hz

ZM

Zygomaticus major: muscle located along the jaw;

generally, a measure of positively valenced emotional

expressivity

xii


Statement of original authorship

The work contained in this thesis has not been previously submitted to meet requirements

for an award at this or any other higher education institution. To the best of my knowledge and

belief, the thesis contains no material previously published or written by another person except

where due reference is made.

Signature: QUT Verified Signature

Date: September 2017

xiii


Acknowledgements

Ever since I came to the realisation that I was drawing near the end of my candidature, one

concern has been at the forefront of my mind: how on earth can I possibly adequately acknowledge

and thank the wonderful people who have supported me?

I’ll give it my best shot, but I may also be buying cupcakes.

I am exceedingly grateful to my principal supervisor, Daniel Johnson, for his continued

reassurance during my moments of anxiousness (of which he is either excellent at sensing or I am

terrible at hiding—probably both) and preternatural ability for guidance. He is an inspiration to

me for his talents as a mentor, researcher, educator and answerer of last-minute panicked emails.

I am also indebted to my associate supervisors: Peta Wyeth, for her encouragement, advice and

keen insight into research approaches; and Simon Smith, who humoured my never-ending

questions about all things psychophysiological with understanding and candour.

I would also like to thank my phenomenal colleagues at the Games Research and

Interaction Design lab—in particular, Mitch, the stats wizard, for making every unit I taught with

him a blast; Kellie, for her wit, cat photos, and encyclopaedic knowledge of research literature;

and Nicole, for kindness, solidarity and yes, also her cat photos. Carody Culver, a big thank you

for your copyediting and proof writing services within such a tight timeframe (and in accordance

with the university-endorsed national ‘Guidelines for editing research theses’) – as well as the

relief it’s brought me in scrolling through my New and Improved Thesis™ with its sense-making

tenses.

I would also like to thank my friends and family. Cilla, Mei, Lyndon, Steph, Hoa, Martin,

Alex, Lall, Vi: thank you for the games, the encouragement and the tolerance of my senseless 4

am messages. Thank you to my parents, Julie and Ray, for your support, cheerleading and

encouragement in my endeavours in education (even as they never seemed to end).

Finally, thank you to my partner, best friend and cornerstone, Cody Phillips—who, as I write

this, has just fallen asleep at 6.00am after keeping me company as I worked through the night. I’d

promise to do the same for you come your thesis submission, but I really like sleeping.

xiv


TABLE OF CONTENTS

1 Introduction ............................................................................................................................ 1

1.1 Background .................................................................................................................... 1

1.2 Research Aim ................................................................................................................. 3

1.3 Contributions .................................................................................................................. 4

1.4 Significance and Scope .................................................................................................. 4

1.5 Thesis Outline................................................................................................................. 6

2 Literature Review ................................................................................................................... 8

2.1 Scope .............................................................................................................................. 8

2.2 Player Experience ........................................................................................................... 9

2.3 Flow and Optimal Psychology ..................................................................................... 10

2.3.1 Challenge and Challenge‒Skill Manipulation ...................................................... 13

2.4 Self-Determination Theory ........................................................................................... 16

2.5 Presence ........................................................................................................................ 17

2.6 Positive and Negative Affect ........................................................................................ 19

2.7 Subjective Methods for Measuring the Player experience ........................................... 20

2.7.1 Survey Method ..................................................................................................... 20

2.7.2 Interviews and Focus Groups ............................................................................... 21

2.7.3 Ethnography, Observation and Think Aloud ........................................................ 21

2.7.4 Game Metrics and Telemetry ............................................................................... 21

2.8 The Psychophysiological Method ................................................................................ 22

2.8.1 Background .......................................................................................................... 22

2.8.2 Benefits and Limitations....................................................................................... 25

2.8.3 Recording ............................................................................................................. 26

2.8.4 Arousal and Valence ............................................................................................ 27

2.8.5 Tonic and Phasic Measurement ............................................................................ 30

2.8.6 Psychophysiological Concepts ............................................................................. 32

2.8.7 Physiological Measures ........................................................................................ 38

2.8.8 Limitations............................................................................................................ 51

2.9 Psychophysiology of the Player Experience ................................................................ 52

2.9.1 Validating the Psychophysiological Method in Games ........................................ 53

2.9.2 Game Effects ........................................................................................................ 57

xv


2.9.3 Social Play ........................................................................................................... 62

2.9.4 Immersion ............................................................................................................ 64

2.9.5 Dynamic Difficulty Adjustment and Biofeedback ............................................... 66

2.9.6 Flow and Challenge ............................................................................................. 68

2.9.7 Gaps Identified in Player Experience Literature .................................................. 72

3 Research Design and Methodologies ................................................................................... 74

3.1 Research Structure and Scope ...................................................................................... 74

3.2 Research Stages ........................................................................................................... 77

3.2.1 Summary Stage 1—Development of Methodologies and Game Artefact ........... 78

3.2.2 Summary Stage 2—Pilot Testing and Design Iteration ....................................... 79

3.2.3 Summary Stage 3—Data Collection and Analysis .............................................. 80

3.2.4 Summary Stage 4—Evaluation ............................................................................ 81

3.3 Stage 1—Development of Methodology and Artefact ................................................. 82

3.3.1 Introduction .......................................................................................................... 82

3.3.2 Theoretical Grounding for Study Design ............................................................. 83

3.3.3 Identifying a Viable Physiological Approach ...................................................... 85

3.3.4 Selection of a Video Game Artefact .................................................................... 93

3.3.5 Design Phase One ................................................................................................ 95

3.3.6 Design Phase One: Artefact Design Flaws .......................................................... 98

3.3.7 Design Phase Two ................................................................................................ 98

3.3.8 Tutorial ............................................................................................................... 102

3.4 Stage 2—Study 1 (Pilot) and Revision ...................................................................... 104

3.4.1 Introduction ........................................................................................................ 104

3.4.2 Methodology ...................................................................................................... 104

3.4.3 Study .................................................................................................................. 105

3.4.4 Revisions to Methodology ................................................................................. 105

3.4.5 Development of Sequencing Software ............................................................... 108

3.5 Stage 3—Study 2 ....................................................................................................... 111

3.5.1 Introduction ........................................................................................................ 111

3.5.2 Methodology ...................................................................................................... 111

3.6 Ethics and Limitations ............................................................................................... 112

3.7 Stage 4—Analysis and Interpretation ........................................................................ 114

xvi


3.7.1 Introduction ........................................................................................................ 114

3.7.2 Scope .................................................................................................................. 114

4 Study 1: Pilot ...................................................................................................................... 116

4.1 Method ....................................................................................................................... 116

4.1.1 Recruitment ........................................................................................................ 116

4.1.2 Measures ............................................................................................................. 117

4.1.3 Laboratory .......................................................................................................... 118

4.1.4 Procedure ............................................................................................................ 118

4.1.5 Participants ......................................................................................................... 120

4.2 Findings ...................................................................................................................... 120

4.3 Discussion .................................................................................................................. 123

4.3.1 Difficulties in Reducing Flow in Immersive Games .......................................... 123

4.3.2 Unsuccessful Condition Design ......................................................................... 124

4.3.3 Challenge‒Skill as an Antecedent ...................................................................... 124

4.3.4 Scale Applicability ............................................................................................. 125

4.4 Conclusions ................................................................................................................ 127

5 Study 2—Main Study ......................................................................................................... 128

5.1 Method ....................................................................................................................... 129

5.1.1 Recruitment ........................................................................................................ 129

5.1.2 Self-Report Measures ......................................................................................... 130

5.1.3 Demographics Questionnaire ............................................................................. 130

5.1.4 Short Flow State Scale ....................................................................................... 130

5.1.5 Player Experience of Needs Satisfaction ............................................................ 130

5.1.6 Intrinsic Motivation Inventory ........................................................................... 132

5.1.7 Psychophysiological Measures ........................................................................... 132

5.1.8 Ethics .................................................................................................................. 133

5.1.9 Procedure ............................................................................................................ 133

5.1.10 Attachment of Psychophysiological Measures ................................................... 135

5.1.11 Electrodermal Activity Attachment ..................................................................... 136

5.1.12 EMG Attachment ................................................................................................ 137

5.1.13 Electrocardiography Attachment ....................................................................... 138

5.1.14 Electroencephalography Attachment ................................................................. 140

xvii


5.1.15 Data Treatment ................................................................................................... 141

5.1.16 Electrodermal Activity ....................................................................................... 142

5.1.17 EMG OO and CS ............................................................................................... 143

5.1.18 Electrocardiography ........................................................................................... 144

5.1.19 EEG .................................................................................................................... 145

5.1.20 Participants ......................................................................................................... 146

5.1.21 Analysis .............................................................................................................. 147

5.2 Self-Report Results .................................................................................................... 148

5.2.1 Confirmation of Optimal and Sub-optimal Conditions ...................................... 148

5.3 Psychophysiological Differences in Optimal and Sub-Optimal Conditions .............. 154

5.3.1 Assumptions and Outliers for Psychophysiological Measures .......................... 154

5.3.2 Results for Electrodermal Activity, High Frequency and Heart Rate ................ 155

5.3.3 Results for Electroencephalography .................................................................. 158

5.3.4 Results for Electromyography - Orbicularis Oculi ............................................ 159

5.3.5 Results for Electromyography - Corrugator Supercilii ...................................... 161

5.3.6 Results for Exploration of Predictive Relationships .......................................... 161

5.4 Discussion .................................................................................................................. 165

5.4.1 Subjective Experience of Play ........................................................................... 165

5.4.2 IMI Interest/Enjoyment ...................................................................................... 165

5.4.3 Flow ................................................................................................................... 167

5.4.4 Player Experience of Needs Satisfaction Competence ...................................... 170

5.4.5 Player Experience of Needs Satisfaction Autonomy ......................................... 172

5.4.6 Player Experience of Needs Satisfaction Presence ............................................ 173

5.4.7 Confirmation of Condition Design Success ....................................................... 174

5.5 Psychophysiological Response to Play ...................................................................... 175

5.5.1 Electrodermal Activity ....................................................................................... 175

5.5.2 Electrocardiography—Heart Rate ...................................................................... 179

5.5.3 Electrocardiography Heart Rate Variability (High-frequency Peaks) ............... 182

5.5.4 Electromyography Orbicularis Oculi ................................................................. 184

5.5.5 Electromyography Corrugator Supercilii ........................................................... 186

5.5.6 Electroencephalography ..................................................................................... 187

5.6 Effect Sizes ................................................................................................................ 193

xviii


6 Summary and Conclusions ................................................................................................. 195

6.1 Self-Report Summary ................................................................................................. 195

6.2 Psychophysiology Summary ...................................................................................... 198

6.3 Exploration of Research Questions and Aim ............................................................. 202

6.4 Applicability of Psychophysiological Assessment ..................................................... 204

6.4.1 Time Costs .......................................................................................................... 204

6.4.2 Viable Psychophysiological Approach ............................................................... 207

6.4.3 Sample Sizes ....................................................................................................... 212

6.4.4 Data Quality Checks ........................................................................................... 213

6.4.5 Automation and Reduction of Participant‒Researcher Interaction .................... 213

6.4.6 Familiarisation with Psychophysiological Principles ......................................... 214

6.4.7 Summary ............................................................................................................ 214

6.5 Limitations and Future Research ................................................................................ 215

6.5.1 Future Approaches to Analysis .......................................................................... 222

6.6 Contributions to Knowledge ...................................................................................... 224

6.7 Conclusion .................................................................................................................. 227

6.8 Final Comments ......................................................................................................... 230

7 References .......................................................................................................................... 231

8 Appendices ......................................................................................................................... 243

8.1 Appendix A—FSS Sample Questions ........................................................................ 243

8.2 Appendix B—Example PENS Items .......................................................................... 244

8.3 Appendix C—IMI Interest/Enjoyment Subscale Items .............................................. 245

8.4 Appendix D—Demographics Questionnaire .............................................................. 246

8.5 Appendix E—Study 2 Script ...................................................................................... 247

8.6 Appendix F—Electroencephalography EEG Outliers ................................................ 251

8.7 Appendix G—Boredom, Balance, and Overload Play Condition Videos .................. 254

1


1 INTRODUCTION

1.1 BACKGROUND

Over the past 15 years, the perception of video games and games culture has transformed

dramatically; while concerns about the worth and potential harms of video game play were

originally the central focus of games research, contemporary research increasingly investigates

the merits of gameplay and the possibilities for promoting positive player experiences (Brand &

Todhunter, 2015; Nacke et al., 2009). As video games have emerged as a leading form of

entertainment, with more than nine in 10 Australian households possessing a gameplay device

(Brand & Todhunter, 2015), so too has the need to explore, evaluate and understand player

experiences.

To accommodate this need, player experience research has become a critical component

of video game evaluation within academic contexts. Founded on user experience, player

experience research allows for the analysis and measurement of video game experiences with a

focus on understanding the relationship and interactions between players and video games (Nacke

et al., 2009). Conceptualising the player experience is undertaken through the lens of

psychological constructs, as informed by psychology and behavioural sciences (Nacke et al.,

2009); these constructs include challenge, flow, immersion/presence, competence, tension and

emotions (Wiemeyer, Nacke, Moser, & Mueller, 2016).

In particular, challenge and challenge‒skill balance (wherein the challenges of the game

are adequately matched by the skills of the player) have been identified, and are widely recognised,

as important elements in ensuring optimal player experiences (Csikszentmihalyi, 1990;

Przybylski, Rigby, and Ryan, 2010; Sweetser & Wyeth, 2005; Fong, Zaleski, & Leach, 2014).

2


Challenge has been established as a key component of games intended to entertain, with player

experience research revealing that players seek, and are driven by, challenge (Andrade, Ramalho,

Gomes, & Corruble, 2006; Lomas, Patel, Forlizzi, & Koedinger, 2013). Furthermore, challenge‒

skill balance has been identified as a probable antecedent to flow, a positive psychology concept

of optimal experience lauded as integral to the development of successful player experiences

(Nakamura & Csikszentmihalyi, 2002; Chen, 2007).

Using a wide range of methods to evaluate the player experience is essential to expanding

understanding of play components such as challenge. Contemporary player experience research

employs a myriad of assessment methods informed by HCI and psychological research contexts,

as well as by commercial games testing spaces (Nacke, Drachen, & Göbel, 2010). Within player

experience research, methodologies have been adapted to assess the unique characteristics of

gameplay (Lankoski & Björk, 2015). In particular, psychophysiological assessment has been

identified as a valuable method for player experience evaluation, and has been employed in a

variety of studies within the player experience domain (Kivikangas et al., 2011). As

psychophysiology offers direct insight into the emotional experience of the player through the

measurement of physiological responses, the method’s relevance to player experience research is

substantial (Nacke, 2013).

Despite the growing popularity of psychophysiological assessment, physiological

responses to player experiences are not yet well understood. While some extant literature has

provided general conclusions about these responses, such as the association of physiological

indicators of mental stress with challenge‒skill balance and flow (Drachen, Nacke, Yannakakis,

& Pedersen, 2010; Keller et al., 2011), other findings have challenged this (Kivikangas, 2006).

Psychophysiological evaluation of like concepts in player experience literature is often mediated

3


by small sample sizes, a lack of uniformity across methodologies and experimental design, or

limitations in the consideration of psychophysiological concepts.

Further psychophysiological research addressing these limitations stands to enhance

understanding of the player experience. Opportunity exists for a rigorous psychophysiological

evaluation of player experience constructs, as informed by psychophysiological concepts and gaps

in player experience literature. Coherent employment of psychophysiological measurements

alongside subjective measurements has been identified as a beneficial approach for the in-depth

analysis of video games (Bernhaupt et al., 2008), and would allow for novel knowledge

contributions to the psychophysiological understanding of the player experience in terms of both

physiological results and the use of psychophysiological measures.

1.2 RESEARCH AIM

This program of research seeks to expand upon current understanding of the

psychophysiological experience of play, as well as the utility and value of psychophysiological

measures as a means of evaluating the player experience. To achieve this, the research undertaken

for this thesis employed a large-scale psychophysiological approach to assessing the player

experience using a variety of psychophysiological measures, a large sample size and a robust and

coherent set of play conditions that allowed for the useful comparison of psychological and

psychophysiological experiences. The research methodology and interpretation of results were

guided by principles established within the psychophysiology field.

4


1.3 CONTRIBUTIONS

The research reported in this thesis specifically investigates constructs of play identified

as critical in promoting optimal player experiences. It is hoped that the research methodologies

and results represent a novel contribution to player experience research.

In particular, the psychophysiological assessment of critical game constructs can provide

games researchers, game developers and game players with a more nuanced understanding of

optimal play experiences. This will allow for more informed evaluation, and therefore

development, of play experiences. These findings may assist burgeoning research investigating

the practical applicability of psychophysiology in spaces such a biofeedback and dynamic

difficulty adjustment (DDA). As psychological constructs employed within games research are

not exclusive to player experience analysis, the results reported here may also prove beneficial to

researchers in other areas, such as user experience, psychology, psychophysiology and pedagogy.

1.4 SIGNIFICANCE AND SCOPE

This research commences with a broad investigation of extant player experience literature,

exploring both critical psychological components of the player experience and the employment of

psychophysiological evaluation to assess these components. A review of psychophysiological

methodologies, measures and principles is also presented. This literature informed the first step in

experimental design through the iterative development of video game conditions designed to act

as a useful means of comparison for psychophysiological results.

An initial study (Study 1) further informed both additional iterations for the game

condition and the experimental design, featuring the introduction of assistive programs for data

5


collection. Upon finalisation of the game conditions and experimental design, the large-scale final

study (Study 2) commenced.

The data gathered from the final study is treated, analysed and interpreted here in

accordance with both player experience and psychophysiological practices. Several

recommendations are then made for the ongoing employment of psychophysiological evaluation

in player experience spaces.

This program of research represents one of the first steps in player experience research

towards the contemporaneous employment of both a wide array of psychophysiological measures

and subjective measures within a large sample size. The results are expected to expand on

understanding of the psychophysiological experience of play, as well as the employment of

psychophysiological evaluation within player experience research.

6


1.5 THESIS OUTLINE

Chapter 2: Literature Review provides an overview of relevant research into both player

experience constructs and psychophysiological assessment. Constructs such as flow, challenge,

presence, self-determination theory (SDT) and enjoyment are investigated, and background in

psychophysiology and psychophysiological principles is provided. A synthesis of literature

employing psychophysiological measures in player experience evaluation is undertaken, allowing

for the identification of gaps in this space; these gaps inform the methodological approach for this

program of research.

Chapter 3: Methodology outlines the research aims and questions developed based on

gaps identified in the literature review. This chapter also details the path taken in experimental

design, discussing the video game conditions, assistive experiment software and chosen

psychophysiological approach; it also gives an overview of both Study 1 and Study 2.

Chapter 4: Study 1 describes the initial non-psychophysiological study undertaken to

review the suitability of the game conditions in the assessment of flow. The results inform several

iterations for the game conditions, methodology and experimental design, and contribute novel

findings about the assessment of optimal and sub-optimal player experiences. Most notably, the

program of research is expanded here to include the investigation of additional player experience

constructs.

Chapter 5: Study 2 details the final large-scale psychophysiological study undertaken to

evaluate optimal and sub-optimal player experiences. Results from both subjective measures

(flow, interest/enjoyment, competence, autonomy and presence) and psychophysiological

measures (EDA, HR, HF Peaks, EMG orbicularis oculi [OO], EMG CS and EEG) are reported

and interpreted.

7


Chapter 6: Summaries and Conclusions provides a summative discussion of the

findings from Study 2, and examines the value and utility of psychophysiology in evaluation of

player experience analysis (as informed by this research). This chapter also details the limitations

and contributions of this program of research, and proposes recommendations for future research.

8


2 LITERATURE REVIEW

2.1 SCOPE

The objective of the literature review in this chapter is to establish understanding of the

player experience of digital games. The literature review concentrates on the review and

exploration of evaluation methods that are employed within player experience research, with

particular focus on the use of psychophysiological assessment.

The chapter begins by broadly exploring the current understanding of the player experience,

and how this is differentiated from the more general ‘user experience’ of non-entertainment

technologies. Key to the player experience are the various subjective psychological constructs

associated with the experience play; the more prominent of these constructs, in terms of their

breadth of study within current literature, are explored and defined here. Additionally, the chapter

identifies measures and models by which these constructs are commonly assessed.

This chapter thus provides a basis for understanding psychophysiology and

psychophysiological assessment through exploring various psychophysiological instruments, their

capacities and the application of the psychophysiological method to player experience analysis.

Relationships between psychophysiological and psychological domains are also explored.

Finally, the chapter discusses methodologies used in prior psychophysiological games

research and identifies gaps in the psychophysiological understanding of the player experience.

These gaps inform the methodologies, aims and research questions of the studies undertaken

within the candidature, as discussed in Chapter 3.

9


2.2 PLAYER EXPERIENCE

Video games represent the fastest growing leisure market in the world (Chatfield, 2010); in

Australia, video games are played by over 68% of the population (Brand & Todhunter, 2015).

Furthermore, 90% of Australian households possess a device used for playing video games (Brand

& Todhunter, 2015). However, despite their popularity, the appeal of video games is deceptively

difficult to explain (Boyle et al., 2012).

The player experience of video games is similar to the standard user experience of software

technology, in that it is defined by what the interaction feels like to the user (Preece et al., 2002)—

specifically in relation to pleasure, fun, enjoyment and aesthetics. Measuring fun is inherently

challenging, as ‘fun’ is not easily quantifiable; this is dissimilar to measuring a program’s

usability, in which the interaction can be determined in terms of efficiency and time (Preece et al.,

2002). Despite its challenges, measuring and defining fun in video game play remains a locus of

player experience research. In an investigation of a uses and gratifications paradigm for video

game play, Sherry et al. (2006) found six dominant dimensions of motivation: arousal (the

stimulation of emotions as a consequence of action), challenge, competition, diversion (the

avoidance of stress or responsibilities), and fantasy (escapism and the ability to do things not

feasible in reality), and social interaction. While these dimensions may not necessarily neatly

equate to a concept of ‘fun’, they do represent a noteworthy step towards understanding and

evaluating video game enjoyment and engagement.

As fun is a defining aspect of the video game player experience, the evaluation or

measurement of fun is especially important. User experience in the context of video games

primarily deals with the user’s emotional response while they interact with the play technology;

for clarity, this thesis refers to the user experience of video games as the ‘player experience’, a

10


term widely used in digital games research literature (Nacke et al., 2009). Wiemeyer, Nacke,

Moser and Mueller (2016) describe player experience as ‘the qualities of the player-game

interactions … typically investigated during and after the interaction with games’. These

experiences are often formally explored within player experience research as constructs of flow,

challenge‒skill balance, affect, presence and motivation, all of which offer an avenue for

investigation in this chapter.

Measuring the player experience can help explain the appeal of games, which can then

facilitate a richer understanding of the medium and contribute to the development of more

successful or appealing video games with improved player experiences. Furthermore, a more

nuanced and informed understanding of psychological experience of play—and the methodologies

used to explore this experience—will help to promote further understanding of motivational and

engagement psychological phenomena.

2.3 FLOW AND OPTIMAL PSYCHOLOGY

Flow describes a mental state characterised by total absorption in and enjoyment of an

activity (Csikszentmihalyi, 1990). Defined by Csikszentmihalyi (1990) as the ‘holistic sensation

that people feel when they act with total involvement’, flow describes an optimal experience

achieved in instances of challenge‒skill balance when engaged in an activity (Nakamura &

Csikszentmihalyi, 2002). For challenge‒skill balance to occur, the demands or challenges of the

task must be met equally by the ability or skill of the individual. In the event of challenge

outstripping skill, the activity becomes too difficult and prevents flow through the introduction of

anxiety; in the event of skill outstripping challenge, the activity becomes too easy and disengages

the individual through apathy or boredom (see Figure 1) (Csikszentmihalyi, 1990). An alternative,

less adverse experience of skill outstripping challenge may also occur during relaxation—while

11


still not allowing for a flow state to occur, relaxation does offer some of the merits associated with

flow (Nakamura & Csikszentmihalyi, 2002). Beyond challenge‒skill balance, an additional

requirement for flow is the presence of clear proximal goals and immediate feedback on the

individual’s progress (Nakamura & Csikszentmihalyi, 2002).

Once these conditions have been met, Nakamura and Csikszentmihalyi (2002) designate the

characteristics of flow as follows:

intense and focused concentration on the task at hand, or the activity taking place in the

present moment

a merging of action and awareness, in which one loses both the consciousness of the self

and awareness of everyday frustrations

loss of reflective self-consciousness—the absence of awareness of oneself as a social

actor, or the loss of concern for others’ perception of the individual

Figure 1. The flow channel. Reproduced from Flow: The Psychology of the Optimal Experience

(Csikszentmihalyi, 1990, p. 74).

12


a sense of control over the activity, and confidence in one’s capacity to deal with the task

at hand

an altered perception of time (distorted temporal experience), often in the sense that time

appears to pass more quickly

an autotelic experience—the experience of the activity at hand is intrinsically motivating

and self-rewarding, such that the activity itself is the reason for the effort expended.

Although the flow construct was originally applied to the experiences of athletes, artists

and chess players conducting their respective tasks, flow has since been expanded to a wide variety

of disciplines and areas, including medicine, education, wellbeing, social activism and writing

(Csikszentmihalyi, 1990; Nakamura & Csikszentmihalyi, 2002). Sherry (2004) has previously

pointed to video games as possessing ideal characteristics for inducing and maintaining flow

experiences. As the concepts of ‘enjoyment’ or ‘fun’ are difficult to define and quantify, the

concept of flow provides a useful lens through which to evaluate the player’s experience of

pleasure during gameplay.

Flow has been formally applied to games by Sweetser and Wyeth (2005) in the development

of the GameFlow model: a set of guidelines, structured by flow, to aid in the design and experience

evaluation of video games. The GameFlow model adapted the characteristics of flow—

concentration, challenge, skills, control, clear goals and feedback—as well as the introduction of

immersion and social interaction in creating a set of guidelines designed to enable or review the

flow experience in video game play. The GameFlow model allowed for Sweetser and Wyeth to

accurately distinguish between high- and low-rated games, and furthermore, to identify the

successes of the former and failings of the latter.

13


Cowley, Charles, Black and Hickey (2008) propose that video games are robustly able to

invoke flow experiences, primarily as a consequence of low investment thresholds (the absence

of financial investment or associated risks) and the opportunity for mastery within the game world.

In particular, Cowley et al. state that game mechanics ascribe to a general and familiar set of rules

that allow for comfortable investment within the game world—in particular, ordered environments

and opportunities for action, advancing in complexity as familiarity with the game also improves,

facilitate ‘immediate access’ to the optimal experience.

Players value video games based on their ability to induce flow experiences (Chen, 2007).

The importance of designing for flow is realised within academic and commercial contexts; Chen

states that ‘most of today’s video games deliberately include and leverage the eight components

of Flow’. Chen proposes that games should try to maintain players within the ‘flow zone’ through

careful manipulation of challenge‒skill balance—ensuring that the player remains within a state

of challenge‒skill balance, and inhibiting the potential for boredom or anxiety to intrude upon the

player experience. Due to the crucial role challenge-skill balance plays in evoking flow, this

component represents one of the primary focuses of the program of research described within this

thesis.

2.3.1 Challenge and Challenge‒Skill Manipulation

Much of game design is rooted in the concept that players seek, and are driven by,

challenge (Lomas, Patel, Forlizzi, & Koedinger, 2013). In Andrade, Ramalho, Gomes and

Corruble’s (2006) survey of the main features of entertaining games, challenge was reported as of

key importance. Successfully completing challenging tasks in game environments generates a

sense of greater self-efficacy and accomplishment for players (Lomas et al., 2013). Research also

suggests that simply undertaking optimally challenging activities—rather than just the experience

14


of succeeding at the task—is an enjoyable experience in itself (Csikszentmihalyi, 1990), which

establishes challenge-based game play as an intrinsically motivating activity. The role of

challenge‒skill balance in ensuring an optimal player experience is further supported by

Przybylski, Rigby and Ryan (2010), who describe challenge‒skill balance as a critical element in

the design and success of arcade games:

The pacing of challenges was designed so players could continually experience enhanced

competence as they progressed in the game, with challenges increasing apace with player

ability. This balancing of game difficulty and player skill was critical to the success of

arcade games; if the challenges underwhelmed players, they would lead to boredom, and if

they overwhelmed the player, they would generate frustration.

The importance of challenge‒skill balance to inducing flow is supported by the research

of Rheinberg and Vollmeyer (2003), who created three game conditions within Roboguard, a

video game in which the player must steer a spaceship out of the path of hostile rockets; these

game conditions varied according to task demands defined by the speed of the attacking rockets

(easy, medium and hard). In alignment with flow literature, the ‘medium’ difficulty was rated

highest for experiences of flow.

Research by Keller and Bless (2008) investigated the experience of flow through the

manipulation of challenge‒skill balance within modified versions of the video game Tetris. In the

program’s first study, three different versions of Tetris were developed: an adaptive condition in

which the difficulty level of Tetris dynamically updated to meet the player’s skill level (achieving

optimal challenge‒skill balance); an overload condition with very high task demands; and a

boredom condition with low task demands. Consistent with the elements of flow, participants in

the adaptive condition reported an altered perception of time (which felt accelerated), greater

involvement and enjoyment, and greater perceived fit of skill and task demands (Keller & Bless,

15


2008). This simultaneously emphasises the crucial role of challenge‒skill balance in seeding the

impetus for flow, and the role flow plays in video game enjoyment.

Despite the importance of challenge and challenge‒skill balance in evoking an optimal

play experience, the notion of challenge is not yet well-defined within games literature (Cox,

Cairns, Shah, & Caroll, 2012). The experience of challenge has been found to depend on both

game genre (Cole, Cairns, & Gillies, 2015) and the relationship of the player with the game or

gaming in general (Alexander, Sear, & Oikonomou, 2013; Lomas et al., 2013). Research

investigating engagement in an educational game found that players were more engaged, and

played longer, when the game presented an easy challenge as opposed to a moderate challenge

(Lomas et al., 2013), complicating the role of challenge‒skill balance as crucial to creating optimal

player experiences. Lomas et al. suggest that a possible explanation for this outcome was the lack

of prior experience among their sample, and the possibility that challenge-seeking behaviours may

only occur after some level of expertise is acquired. This is supported by other research that

discovered ‘casual’ players of a game gained more enjoyment from easier difficulties, regardless

of their aptitude in the game, whereas ‘experienced’ players’ game enjoyment was predicated upon

challenge (Alexander et al., 2013). This finding has been reflected in the evaluation of flow

experiences, in which increased challenge in game tasks was found to both induce greater flow

experiences for highly skilled players, and reduce flow experiences for low- and moderately-

skilled players (Jin, 2011). Cox et al. (2012) suggest that one mediator of this experience may be

the player’s self-perception: if they do not perceive themselves as possessing the requisite

expertise for overcoming the game challenges, the positive experience of the game—explored as

immersion within their research—will be reduced. It is possible that these findings, explored

further, may provide an explanation for some subgroups’ (e.g. older players) reticence to engage

in certain video game contexts, such as online multiplayer.

16


These findings—the effects of perceived challenge, and variance of player skill—have

prompted research questioning the centrality of challenge‒skill balance in invoking flow

experiences. In a meta-analysis of 28 studies, Fong, Zaleski and Leach (2014) investigated the

role of challenge‒skill balance in both flow and intrinsic motivation within a variety of contexts.

Their results revealed a moderate relationship between challenge‒skill balance and flow, but the

authors maintain that challenge‒skill balance remains a ‘robust contributor’ to flow alongside

‘clear goals’ and ‘sense of control’ (the other antecedents of flow identified by Csikszentmihalyi

[1990]). Furthermore, Fong et al. suggest that the difficulty of operationalising challenge may

explain the ‘moderate’ result. Ultimately, the authors conclude that challenge‒skill balance

remains robustly related to feelings of flow or optimal experience.

2.4 SELF-DETERMINATION THEORY

Motivation and need satisfaction in video games has been understood through the

application of SDT, an established psychological theory of motivation concerned with the

fulfilment of a universal need to experience competence, relatedness and autonomy. SDT

primarily addresses factors that enable intrinsic motivation, which is a core reason for involvement

in play and sport (Ryan, Rigby, & Przybylski, 2006). As self-determination theory is not one of

the core focuses of this thesis, but instead peripheral to interpreting results, this section will be

feature broader—rather than in-detailed—discussion for the benefit of contextualisation.

Autonomy refers to the sense of volition or willingness experienced when doing a task

(Deci & Ryan, 2000), and is positively associated with increased intrinsic motivation. Enabling a

sense of freedom and choice for players, providing opportunities that interest them and avoiding

factors that diminish volition (such as controlling or limiting tasks) promotes autonomy, in turn

promoting willingness to engage in the activity (Ryan et al., 2006).

17


Competence addresses feelings of effectiveness and a need for challenge within the game

(Ryan et al., 2006). To achieve satisfied competence, players should be optimally challenged,

receive positive feedback and be presented with the opportunity to improve on their abilities and

skills.

Relatedness investigates feelings of connection to others. Within gaming, this primarily

pertains to experiences of relatedness to human others in multiplayer games; however, this may

also address the sense of connection to computer-controlled non-player characters (NPCs) (Ryan

et al., 2006). Research conducted by Coulson, Barnett, Ferguson and Gould (2012) offers credence

to the latter concept with its finding that players form ‘authentic emotional attachments’ to NPCs

that arise as a consequence of NPC appearance, friendliness and utility.

Studies of SDT in video game contexts have revealed less competence and autonomy

when players engage in co-located play with others than in solo play or online play (Johnson,

Wyeth, Clark, & Watling, 2015), autonomy, competence and relatedness as positive predictors of

game enjoyment and future play, and positive associations between autonomy and competence

and post-play mood (Ryan et al., 2006).

2.5 PRESENCE

Within player experience research, presence is often divided into two forms: social

presence, or the feeling of being with another person within a mediated world (for example, within

a chatroom) (Biocca, Harms, & Burgoon, 2003), and spatial presence, which describes the

experience of being physically located within the mediated world (Wirth et al., 2007). Spatial

presence, facilitated by engaging narrative and visually pleasing aesthetics, is characterised by the

International Society of Presence Research as ‘[occurring] when part or all of a person’s

18


perception fails to accurately acknowledge the role of technology that makes it appear that s/he is

in a physical location and environment different from her/his actual location and environment in

the physical world’ (International Society for Presence Research, 2000). As with the discussion of

Self-Determination Theory in the previous section, the following discussion of presence will only

be broadly discussed.

The concept of presence is often conflated with the concept of immersion within player

experience research. Vella (2016) distinguishes between the two constructs, establishing that

while presence relates more to the sense of physical location within the game world, immersion

describes engagement over time (for example, emotional involvement in the game narrative).

Vella further suggests that game experiences lacking engaging game worlds, such as Tetris, may

be simultaneously capable of inhibiting presence while evoking immersion, whereas games that

do feature game worlds, such as BioShock (a dystopian first-person shooter featuring realistic and

well-crafted environments), are capable of promoting both experiences.

In player experience assessment, the PENS scale evaluates both immersion and presence as

a single construct comprising items measuring physical, emotional and narrative presence. The

conflation of these constructs is undertaken as an investigation of the ‘illusion of non-mediation’,

wherein the player perceives themselves as engaged and present within a mediated world

(Lombard & Ditton, 1997).

Presence is positively related to space exploration and the need for discovery within game

worlds (Skalski, Dalisay, Kushin, & Liu, 2012), and has been identified as a key contributor to

greater enjoyment of games (Przybylski et al., 2011; Lombard & Ditton, 1997). Games played at

higher difficulties (that is, games that have not been identified as ‘easy’) or in first-person view,

19


or that feature realistic environments, have also been identified as more likely to provoke feelings

of presence (Ravaja et al., 2004).

2.6 POSITIVE AND NEGATIVE AFFECT

Affect is an experiential phenomena concerned with state feelings or emotions. One of the

broadest measures of the player experience, affect evaluates the player’s positive or negative

feelings. There is some ambiguity in the literature concerning ‘mood’ and ‘affect’—while the

terms are often used interchangeably, such as in Ravaja et al.’s study of phasic emotional responses

(Ravaja, Saari, Salminen, Laarni, & Kallinen, 2006a), an important distinction exists between the

two. While affective states are emotions experienced during an event (or, in the context of this

research topic, within player experience), mood is a lasting disposition that may affect a person’s

overall perception of an event (Sims, 1988). The glossary of the American Psychiatric Association

states that ‘affect is momentary (like weather), while, mood is a prolonged emotion (like climate)’;

in the context of human emotion, “affect” may be a transitory or reactionary emotion (e.g.

surprise), and mood a state emotion (e.g., happiness or contentment). Within the context of player

experience evaluation, it is necessary to measure affective states, as these are the emotions

experienced during play (Mandryk & Atkins, 2007). As with the preceding discussions of Self-

Determination Theory and presence, the following section will again approach this concept

broadly.

Positive affect has been linked with positive player experiences, such as competence

satisfaction (Ryan et al., 2006), convergence between the player’s experiences of self and the

player’s ideal self during play (Przybylski, Weinstein, Murayama, Lynch, & Ryan, 2012) and

prosocial play (Saleem, Anderson, & Gentile, 2012). Conversely, a study by Jennett et al. (2008)

examined the relationship between fast-paced games, negative affect and state anxiety; while they

20


found that negative affect and state anxiety were higher for faster paced games than slower games,

the results were not significant. Negative affect and emotions have thus been recommended for

further study (Jennett et al., 2008; Boyle, Connolly, Hainey, & Boyle, 2012).

2.7 SUBJECTIVE METHODS FOR MEASURING THE PLAYER EXPERIENCE

Player experience research employs a number of methods for experience evaluation,

informed by both HCI research contexts and commercial games testing (Nacke, Drachen, &

Göbel, 2010). In commercial spaces, play experience evaluation has historically been informal

and performed with in-house testers; however, contemporary commercial play-testing employs

formalised strategies from HCI research and usability fields (Nacke, Drachen, & Göbel, 2010).

Within player experience research, research methodologies have been adapted for assessing the

unique characteristics of gameplay (Lankoski & Björk, 2015). Several evaluation methods used

for player experience research within both academic and commercial contexts are discussed in this

section.

2.7.1 Survey Method

As a form of structured interview, surveys are one of the most frequently used research

methods for subjectively measuring the player experience (Cote & Raz, 2015). The widespread

use of surveys in player experience research is partly owed to their ease of distribution and flexible

design. In industry and research settings, a post-play questionnaire can be delivered to the

participant sample, allowing for a mix of quantitative and qualitative responses relating to the play

experience. Rich qualitative data about the player experience can be captured through open-ended

questions, which offer insight into the broad themes and trends that emerge (Cote & Raz, 2015).

Quantitative data is typically captured through the inclusion of Likert scale items, which limit the

scope of participant responses, since participant experiences are numerically signified by their

21


level of agreement with a statement. In the context of player experience evaluation, constructs

relating to player enjoyment are particularly examined (Nacke, Drachen, & Göbel, 2010).

2.7.2 Interviews and Focus Groups

Like the survey method, interviews and focus groups allow for the collection of subjective

player experience data. Unlike the survey method, however, interviews and focus groups centre

on the collection of qualitative data, are typically semi-structured and make extensive use of open-

ended questions, thus allowing researchers to probe participants’ responses for a more detailed

and nuanced understanding of player experience phenomena (Cote & Raz, 2015). Qualitative data

collected through interviews and focus groups are often used to triangulate experiential

phenomena between multiple data streams in mixed-methods research (Cote & Raz, 2015).

2.7.3 Ethnography, Observation and Think Aloud

If participants’ in-game behaviour forms part of the research question, an ethnographic-

or observation-driven approach to examining the phenomena may occur. Observing participants

as they play a game allows for insight into the player experience that might otherwise be

uncaptured by the game’s telemetry—for example, player behaviours or participants’ cognitive

load (Lieberoth & Roepstorff, 2015). A limitation of this approach is that it relies on the

researcher’s subjective observation of events about the player’s interaction with the game.

Observing participants in laboratory settings is often coupled with a think-aloud exercise, so that

participants vocalise their thought process, giving insight into what guides their behaviour.

2.7.4 Game Metrics and Telemetry

The digital nature of video games allows for the collection of vast amounts of

instrumentation data. By logging and tracking in-game events such as input commands,

interactions with entities, movement through the game space and so on, researchers are able to

22


determine the types of interactions that occur in the game (Nacke, Drachen, & Göbel, 2010). While

this approach yields valuable insight into the events taking place during play, a core limitation of

this method is that interpreting game metrics is difficult and largely requires subjective

interpretation of user stories, as well as the development of player personas to provide structure to

the data investigation (Tychsen & Canossa, 2008).

2.8 THE PSYCHOPHYSIOLOGICAL METHOD

‘While game metrics provide excellent methods and techniques to infer behavior from the

interaction of the player in the virtual game world, they cannot infer or see emotional signals of a

player.’ (Nacke, 2013, p. 585)

The psychophysiological method allows for the covert, direct interpretation of player

emotional response to play experiences through physiological measurement (Nacke, 2013). The

coherent employment of psychophysiological measurements alongside subjective measures has

been identified as a beneficial approach for in-depth analysis of the player experience (Bernhaupt

et al., 2008), and has recently enjoyed growing popularity within player experience research and

industry. The following section explores the psychophysiological method, constructs and

measures, and seeks to establish a broad understanding of psychophysiology for employment in

player experience evaluation.

2.8.1 Background

Andreassi (2007, p. 2) defines the field of psychophysiology as ‘the study of relations

between psychological manipulations and resulting physiological responses, measured in the

living organism, to promote understanding of the relation between mental and bodily processes’.

By this definition, ‘mental processes’ encompass both emotional responses (such as fear, anger or

23


joy) and cognitive processing (such as problem-solving); the use of psychophysiological

instruments allows for a researcher to record and identify physiological responses associated with

these behaviours. These physiological responses are sourced from a multitude of sites on or within

the body. A characteristic example of a physiological response to emotion, for example, is an

increased heart rate (HR) associated with stress or fear; similarly, problem-solving is also

associated with an increase in cognitive activity, as recorded by sensors monitoring the brain

(Melillo, Bracale, & Pecchia, 2011; Stern, Ray, & Quigley, 2001).

Psychophysiological signals originate as electrochemical changes in neurons, muscles and

gland cells (Stern et al., 2001, p. 33), as controlled by the nervous system (see Figure 2). The

nervous system is divisible into the central nervous system (CNS), which consists of the brain and

Figure 2. Divisions of the nervous system.

.

24


spinal cord, and the peripheral nervous system, which consists of nerves and ganglia (nerve cell

clusters) outside the brain and spinal cord (Andreassi, 2007, pp. 11‒12). Furthermore, the

peripheral nervous system contains the autonomic nervous system (ANS), sub-divisible into the

parasympathetic nervous system (PNS) and sympathetic nervous system (SNS), which are

dominant in situations requiring rest and ‘work’ (that is, energy mobilisation), respectively. The

ANS is responsible for many of the physiological signals of interest in the field of

psychophysiology—for example, eccrine sweat gland activity is controlled by the SNS (a sub-

division of the ANS), and is a widespread measure of psychophysiological arousal (Andreassi,

2007, p. 13).

Unfiltered physiological signals are obtained through the placement of electrodes on

human skin, the locational site of which depends on the specific psychophysiological measure

being used. For example, a measure reliant on muscle activation—such as electromyography

(EMG)—requires that the electrodes be placed on the specific muscle site being investigated, such

as the zygomaticus major (ZM: a cheek muscle associated with positive emotion). These signals

are simply positive or negative voltage, characterised by amplitude, latency and frequency (Nacke,

2013). For analysis, the signals are then processed and filtered in accordance with the correct

procedures for the specific measure chosen.

Psychophysiology has been used primarily in the fields of medicine, neurobiology,

psychology and, most recently, HCI (Nacke, Grimshaw, & Lindley, 2010). Wastell and Newman

(1996) measured HR and blood pressure (BP) to determine the stress of British ambulance

dispatchers in the switch from paper to digital systems; Ward and Marsden (2003) collected skin

conductance, blood volume pulse and heart rates to gauge the user experience of navigating

websites of varying design quality; and Wilson and Sasse (2000) evaluated psychophysiological

25


response to video and audio degradations. The success and widespread adaptation of

psychophysiological measures in other fields lends credibility to their use in a video game-specific

context, particularly given their frequency in HCI research (Nacke et al., 2010b).

2.8.2 Benefits and Limitations

Whereas psychophysiology historically required the use of cumbersome equipment and

manual recording, modern technological advances have allowed for relatively unobtrusive

psychophysiological insight into the typical healthy user (Cacioppo, Tassinary, & Berntson, 2000,

p. 3). As physiological responses are generally uncontrollable or unmodifiable by the participant,

and occur spontaneously in response to stimuli, the psychophysiological method is an effective

objective approach for experience analysis (Nacke, 2013). The continuous recording of

physiological signals also allows for uninterrupted data collection, removing the necessity to pause

an experiment for the distribution of a measure that requires active participant interaction.

Despite these advantages, the complexities associated with recording, treating and

analysing psychophysiological data often present as a barrier to its use in non-clinical contexts.

The instruments themselves can often be invasive or uncomfortable for the participants, detracting

from naturalised settings and environments. These obstacles are being resolved by recent

developments in consumer-grade psychophysiological equipment, with the introduction of more

intuitive and less conspicuous products such as the single-sensor MindWave EEG headset

(NeuroSky, 2015) and portable Empatica wristband (Empatica, 2017). However, research has

shown that consumer-grade products often exhibit high variability in performance (Maskeliunas,

Damasevicius, Martisius, & Vasiljevas, 2016; Duvinage et al., 2013), suggesting that a middle

ground may be preferable for academic and industry research.

26


2.8.3 Recording

2.8.3.1 Electrodes and Preparation

To obtain the physiological response, electrodes are employed to collect the bioelectric

potentials of the site—for example, muscle or gland—being recorded. This information is then

relayed to a computer or polygraph, where the signal is filtered and amplified (Stern et al., 2001).

The data analysis then generally occurs as a separate event.

Typically, electrodes are placed on the surface of the skin (cutaneous electrodes), although

subcutaneous or needle electrodes can also be used (Stern et al., 2001, p. 36). These cutaneous

electrodes, in the form of metal discs, are placed in pairs to allow for a path to be established

between each electrode. The electrodes are aided by a conductive gel or paste, of which the

purposes are threefold: the gel or paste is often inserted into a socket between the electrode itself

and the skin (providing a cushion that reduces risk of disturbances otherwise possible in direct

electrode-to-skin contact, such as movement artefacts caused by small movements); it lowers the

impedance between the electrode and the skin; and finally, the gel can act as an adhesive that helps

affix and maintain the electrode to and in the correct position (Stern et al., 2001, p. 38). A plastic

adhesive collar attached to the electrode also affixes the electrode to the skin, and often, tape is

securely placed over the attached electrode to prevent detachment or slipping.

As the outer layer of skin is largely composed of dead skin cells, various oils and dirt, the

site of electrode placement is prepared to ensure minimal impedance and an uncompromised

signal. To do this, skin is often gently abraded until it becomes red with an abrasive cloth and/or

skin preparation gel. The location is then wiped with alcohol and water to ensure a clean surface,

except in the instance of electrodermal activity (EDA) (Stern et al., 2001, pp. 38–39). Finally, the

27


resistance between the applied electrodes is checked with an impedance meter. An impedance of

< 10 kΩ is recommended (Stern et al., 2001, p. 39).

2.8.3.2 Filtering

The signal acquired by the electrodes is filtered during both amplification and pre-analysis

to attenuate the frequencies, so that any data above or below a certain threshold is allowed to pass

(Stern et al., 2001, p. 41). For example, a low-pass filter may be set to discard any information

received below 12 Hz; similarly, a high-pass filter may be set to discard any information above 12

Hz. This is useful for noise reduction, such as movement artefacts or interference from co-located

physiological responses that occur at a greater or lower range than the target physiological

measure.

A band-pass filter, a combination of both the high-pass and low-pass filters, allows for

only a range of frequencies to pass. This is beneficial should the researcher or clinician be

interested in a specific physiological response that only occurs within a certain range. For example,

a beta wave collected in electroencephalography (EEG) is only present between 13 and 30 HZ—

as such, a band-pass filter can attenuate any frequencies received above or below this range. A

notch or band-reject filter is set to the frequency of the AC current in the country in which the

information is collected, allowing for the attenuation of interfering frequencies (Stern et al., 2001,

p. 41)—in Australia, it is set to 50 Hz in congruence with international voltage standards.

2.8.4 Arousal and Valence

Within the overlapping fields of player experience and emotional psychology, Russell’s

dimensional theory of valence and arousal is particularly prominent (Mandryk et al., 2006b;

Nacke, 2013). Russell (1980) suggests that emotion is not rigid, and supports a dimensional theory

28


of emotion: that all emotions can be located in a two-dimensional space, as coordinates of valence

and arousal. Figure 3 provides an example of Russell’s circumplex model of affect.

Russell’s two-dimensional space of valence and arousal has been commonly used in

psychophysiological studies (Mandryk et al., 2006a, 2006b & 2007; Nacke et al., 2010a & 2010b;

and Lang, 1995). This is likely due to the limitations of psychophysiological responses as a

measurement of explicit emotions, necessitating a reliance on valence and arousal feedback

(present in the axes of Russell’s model).

Figure 3. Russell's Two-Dimensional Model of Emotion (reproduced from Russell, 1980).

29


Figure 4. The relation of valence and arousal (reproduced from Ambinder, 2011).

As psychophysiological instruments are unable to identify specific emotions (Stern et al.,

2001), psychophysiological evaluations of the player experience instead measure levels of valence

and/or arousal. Valence (i.e., tone) reflects the positivity or negativity of an emotional

experience—for example, whether the user is experiencing happiness or sadness. Arousal (i.e.,

activation) reflects the level, or extremity, of the valence—for example, high arousal and high

valence could reflect intense happiness, whereas low valence, high arousal could reflect intense

sadness (Ambinder, 2011; Nacke, 2010b; Ravaja, 2006a). A concise representational interplay of

valence and arousal is provided in Figure 4. Psychophysiological analysis is often restricted to

measurements of arousal and valence; as a result, multiple measurements are often used

concurrently to aide in triangulating the correct emotion (Nacke, 2013). This is aided by the

inclusion of survey and interview methods (Nacke, 2013).

Non-dimensional models of affect—that is, the categorical treatment of emotions as

distinct and fundamentally discrete entities—are also of note. Ekman (1992) identifies six basic

universal emotions, experienced to varying degrees of intensity: anger, fear, sadness, enjoyment,

30


disgust, and surprise. While these emotions were identified in a cross-cultural examination of

facial expression, Ekman (1992) also proposes that these emotions differ in behavioural and

physiological response; furthermore, that these emotions each represent a ‘family’ of related states

that share commonality in expression and physiology (e.g. there are different types of ‘anger’, but

all share the commonality of constricted brow muscle movement or pursed lips).

It should be noted that physiological arousal does not simply extend on a unidimensional

continuum ranging from low to high activation, as evident in the principle of stimulus-response

specificity (Stern et al., 2001, p. 53; see section 2.8.6.4). Arousal can present in the cortical (e.g.,

alpha band EEG frequency), sympathetic (e.g., HR and BP) and somatic (e.g., muscle tension)

systems (Bradley, 2000, p. 604), and as such, does not necessarily increase uniformly across all

systems (Lacey, 1967). This is classically represented in the flight, fight or freeze responses,

particularly concerning somatic response. Despite a high level of arousal, the somatic system may

either incur a high level of activation (muscle activity in the flight or fight response) or a relatively

low level of activation (as in the freeze response). (See section 2.8.6.4 for extrapolation on these

concepts.)

2.8.5 Tonic and Phasic Measurement

Physiological activity is typically analysed in terms of two domains: tonic and phasic.

Tonic and phasic activity are the windows of physiological response that are investigated in

psychophysiological research. (Stern et al., 2001, p. 47), and the type of activity analysed—tonic

or phasic—depends on the aims of the study. Tonic and phasic psychophysiological approaches

to player experience evaluation permit different perspectives of analysis. A tonic approach

provides overall emotional feedback, allowing researchers to gain an overview or ‘average’ of the

31


experience; alternatively, a phasic approach enables researchers to look at individual events (and

their evoked responses) within an experience (Ravaja et al., 2006a; Mandryk et al., 2006b).

Tonic Activity: The purpose of tonic activity analysis is twofold. Tonic activity is

considered to be the background, or baseline level, of activity (Stern et al., 2001, p. 49). It is thus

collected over variable time intervals (from 30 seconds to several minutes, depending on the

physiological measure being recorded) as a baseline prior to exposure to a stimulus. This baseline

activity is recorded during participant inactivity, during which the participant is generally asked

to relax. This allows for a useful point of comparison in analysis, enabling differentiation between

the participant’s natural or resting physiological state and their reaction to stimulus, as in phasic

analysis. The second purpose of tonic activity allows for ongoing analysis of continued exposure

to a stimulus or experience. This generates an ‘average’ physiological experience associated with

the stimulus—for example, a participant’s average HR throughout a 10-minute hazardous driving

experience, compared with the participant’s average HR in a non-hazardous driving experience or

when at rest.

Phasic Analysis: Phasic analysis allows for granular investigation of discrete

physiological response to specific events or stimuli (Stern et al., 2001, p. 50). As in the hazardous

driving example, phasic analysis in this scenario is useful for evaluating the participant’s

immediate response to a specific hazard in the experiment condition (as opposed to the average

response to the overall condition). This response may be an increase or decrease in amplitude or

frequency of the response—for example, a five-second spike in skin conductance levels

immediately following exposure to a stimulus.

32


The method of psychological analysis is dependent on the aims of the research program.

Neither is phasic or tonic analysis mutually exclusive; both the average response and discrete

responses can be useful when evaluating the psychophysiological experience.

2.8.6 Psychophysiological Concepts

In evaluating psychophysiological response, researchers should be aware of certain

principles of psychophysiology that may influence interpretation of results (Stern et al., 2001, p.

52). Such an understanding will improve accuracy in data collection and analysis, and allow for

informed and critical reading of extant psychophysiological literature within both

psychophysiological and player experience research.

2.8.6.1 Habituation

Habituation is described by Stern (2001, p. 54) as a concept ‘as basic to psychophysiology

as the concept of arousal’. Habituation describes the decrease in psychophysiological

responsiveness that occurs with exposure—specifically, repeated exposure—to the same stimulus

or set of stimuli (Andreassi, 2007, p. 353). Andreassi (2007, p. 353) provides the example of

changes in EDA that occur when a person’s name is called; if the name is repeated over and over

again, the novelty diminishes, and so too does the electrodermal response. Öhmann, Hamm &

Hugdahl et al. (2000, p. 540) describe exposure to a single intense stimulus, wherein electrodermal

and HR components were reduced within 10 trials of repeated exposure to the stimulus. The

presence of habituation is ubiquitous across a range of physiological measures, and may often be

a function of repetition-induced boredom (Stern et al., 2001, p. 55; Andreassi et al., p. 353).

The presence of habituation is commonly observed in, and has implications for, both

short-term and longitudinal psychophysiological assessment (Stern et al., 2001, p. 55). Habituated

response can occur within a single testing session, but generally occurs slowly in response to

33


particularly intense, complex or unique stimuli (Stern et al., 2001, p. 55); in terms of player

experience research, games may present a complex enough stimulus to warrant this. Stern et al.

(2001, p. 55) also suggests that habituation may be minimised by interruption, either through the

requirement of a response (e.g., a survey response) or the introduction of novel stimuli. Despite

this, the ubiquity of habituation necessitates its consideration in experiment design and results

interpretation.

2.8.6.2 Orienting

The orienting response describes a change in physiological response that occurs as a

consequence of exposure to novel stimuli (Öhmann, Hamm, & Hugdahlm 2000, p. 542). Stern et

al. (2001) detail some of the major components of orienting response as increased EDA, delayed

respiration followed by an increase in frequency and amplitude and HR deceleration. Because the

orienting response is sensitive to novel stimuli, it habituates rapidly after continued exposure to

the stimuli (Stern et al., 2001); in player experience analysis, the orienting response may be

circumvented with familiarisation of the game stimulus prior to data collection.

2.8.6.3 Startle Response

Startle response is the physiological response elicited by the abrupt introduction of an

intense stimulus; Stern et al. (2001) provide the example of a sudden thunder strike. Reflexive eye

blinking, HR acceleration and rapid habituation of the physiological response immediately

following are typical of startle response. To minimise the risks of startle response influencing data,

Stern et al. (2001) recommend discarding data from initial exposure to a stimulus; the risk for

unexpected external stimuli, such as a door slamming within the building, should also be

minimised. As video games often feature the rapid introduction of confronting or intense stimuli,

there is a potential for startle response in play experiences; Kivikangas (2006) identifies the chance

34


of startle response influencing facial EMG results in their psychophysiological assessment of

video game flow.

2.8.6.4 Stimulus-Response Specificity and Directional Fractionation

Physiological activation does not occur along a single unidimensional continuum. Lacey

(1967) proposes three complex forms of arousal: cortical, autonomic and behavioural.

Furthermore, Lacey states that one form of arousal cannot consistently be used as a valid measure

of another. Stimulus-response specificity refers to a pattern of physiological response to a specific

stimulus (Andreassi, 2007, p. 345). This pattern of physiological results may be interpretable as

the pattern of response associated with specific emotional states (e.g., fear, disgust or happiness)

(Andreassi, 2007, p. 345). For example, Stern et al. (2001) provide the example of a missing

wallet: noticing a missing wallet (or any missing valuable object) may prompt a specific pattern

of physiological response, such as an increase in muscle tension alongside a decrease in heart and

respiration rate.

This divergence in physiological response (a decrease in heart and respiration rate

alongside an increase in muscle tension) is referred to as directional fractionation, and contradicts

the notion that the ANS response must covary in a single direction (increase or decrease) in

response to stimuli (Andreassi, 2007, p. 345). Stern et al. (2001) also give an example of a soldier

on guard duty: upon hearing an unidentified noise, the soldier may experience an increase in

cortical arousal, increase in EDA and a decrease in HR. Increased or decreased activation in one

measure—otherwise interpreted as increased or decreased arousal—should thus not be considered

reflective of the whole experience; in the case of the soldier, the interpretation of HR alone may

not allow for the interpretation of a psychologically arousing experience.

35


2.8.6.5 The Psychological and Physiological Domains

It is important to consider that the nature of psychophysiological response is such that a

single physiological response is not necessarily singularly and uniquely related to a single

psychological response, insofar as a single psychological state, or discrete emotional response,

exists (Nacke, 2013). Cacioppo et al. (2000, p. 12) identify five possible relationships between the

psychological and physiological domains (see Figure 5):

Figure 5. Relationships between psychological (ψ) and physiological

(ϕ) domains. Reproduced from Handbook of Psychophysiology

(Cacioppo, Bernston, & Tassinary, 2000, p. 12).

36


One-to-One: A single physiological response is directly associated with a single

psychological response, and vice versa. This allows the researcher to determine a specific mental

process or psychological element based solely upon a physiological response. Such a relationship

is rare.

One-to-Many: A one-to-many relationship, suggesting that a single psychological

response is associated with a subset or multitude of physiological responses. This can often be

simplified to a one-to-one relationship, as the identification of a particular set of physiological

responses may be robustly associated with a single psychological response. In this instance, the

numerous physiological responses are regarded as one set or group, or a single unified

physiological response to a mental event. As with the one-to-one relationship, this is uncommon.

Many-to-One: The inverse of the one-to-many relationship. The many-to-one

relationship is the association of a single physiological response with a multitude of psychological

responses; this relationship is the one most often examined in psychophysiological research

(Nacke, 2013).

Many-to-Many: A many-to-many relationship suggests that two or more psychological

responses are associated with same subset (multitude) of physiological responses; it thus does not

allow for the inference of which psychological process the physiological signals may be

responding to.

Null: No association between physiological and psychological response.

The first (one-to-one) and third (many-to-one) relationships are those most commonly

examined in psychophysiological research, as they allow for an assumption about the

psychological state to be made based on physiological response (Cacioppo et al., 2000, p. 12).

37


Physiological responses are thus better understood as ‘elements of sets with fuzzy boundaries’

(Nacke, 2013).

The approach to emotion analysis discussed within this program of research, and within

much psychophysiological literature, is that of an objectivist approach: the treatment of emotion

as objective psychological concepts. However, an alternative constructivist approach is one that

examines emotion as an interaction informed by and interpreted through cultural and social

experiences (Boehner, DePaula, Dourish, & Sengers, 2007), which are not typically assessed in

HCI environments.

38


2.8.7 Physiological Measures

The next section discusses the various physiological measures employed in

psychophysiological research (see Figure 6). These measures record physiological activity from

the central, peripheral, somatic and autonomic nervous systems through the use of cutaneous

electrodes, infrared sensors and optical sensors. The general applications of these measures are

then detailed.

Figure 6. Illustrations of various phasically occurring physiological measures. Reprinted from Handbook of

Psychophysiology (Cacioppo, Bernston, & Tassinary, 2000, p. 706).

39


2.8.7.1 Electrodermal Activity

EDA—previously known as galvanic skin response (GSR)—is one of the most widely

used measures of psychophysiology, owing largely to its relative ease of deployment and

quantification, as well as its sensitivity to psychological stimuli (Dawson, Schell, & Filion, 2000,

p. 200). Primarily a measure of arousal, EDA is the study of electrical activity of the skin, as

recorded by cutaneous electrodes. The activity is generated by an interaction between the SNS and

local processes in the skin (Boucsein, 1992). Two fundamental methods are used for EDA: the

exosomatic method and the endosomatic method. The exosomatic method records skin

conductance and resistance levels, whereas the endosomatic method records skin potential

response. As the exosomatic method is the one chiefly used in contemporary research (Dawson et

al., 2000), this is the one discussed in this section.

Skin is a protective barrier that, among other functions, aids in bodily temperature control

through follicle dilation and sweating. The measurement of EDA in psychophysiology is the

investigation of the psychological states and processes that provoke sweating (Dawson et al.,

2000) through recording sweat gland activity originating in the dermis.

Within the dermis are two types of sweat glands: apocrine and eccrine. Apocrine sweat

glands are large, terminate in hair follicles and are found in the genital and armpit areas (Andreassi,

2007, p. 315); while their role in human physiology is not well understood, they have been

associated with the production of pheromones in animals (Dawson et al., 2000). Conversely,

eccrine sweat glands are distributed generally over most of the human body surface and are found

in their highest concentrations on the palms of the hands (palmar region) and the soles of the feet

(plantar region) (Andreassi, 2007, p. 316). While the primary function of most eccrine sweat

glands is that of thermoregulation, those found on the palmar and plantar sites are more strongly

40


associated with ‘grasping behaviour’ and are thus more responsive to emotional rather than

thermal stimuli (Dawson et al., 2000). This is most noticeable in the phenomenon of ‘clammy

hands’ in stressful or adverse contexts, regardless of the external temperature.

Despite its possible adaptive origins, increased EDA is not solely associated with low-

valence emotions such as stress or anxiety. EDA is responsive to a breadth of stimuli, including

‘stimulus novelty, surprisingness, intensity, emotional content, and significance’ (Dawson et al.,

2000, p. 212). This stimuli sensitivity emphasises the necessity for a controlled experimental

paradigm to restrict EDA response to the investigated variables (although spontaneous activity

remains challenging to control against). The recommendation is thus to change only one aspect of

a stimulus at a time (Dawson et al., 2000). The assistive use of alternative analysis methods, such

as survey and interview, can also aid in

determining the stimuli behind

fluctuations in EDA.

For optimal recording of

eccrine sweat gland activity, EDA

electrodes are placed on either palmar

or plantar sites. The recommended

palmar sites are the medial phalanges,

distal phalanges and hypothenar and

thenar eminences of the palm (see

Figure 7). The distal phalange site has been

shown to have the greatest concentration

of eccrine sweat glands, and is thus the

Figure 7. Optimal placement of EDA electrodes on palmar

sites. Reprinted from Handbook of Psychophysiology

(Cacioppo, Bernston, & Tassinary, 2000, p. 205).

41


recommended placement site for electrodes, unless experiment requirements or participant injury

on the site prevents it (Dawson et al., 2000).

2.8.7.2 Electromyography

EMG measures the electrical potential of muscle activation (contraction) following neural

stimulation (Stern et al., 2001), through either cutaneous electrodes or subcutaneous needle

electrodes (the latter used more commonly in medical contexts) (Andreassi, 2007, p. 283). As with

most psychophysiological methods, EMG can be employed to investigate a variety of topics, such

as motor performance, cognition, sleep and emotional expression (Andreassi, 2007, p. 277)—the

latter topic is the focus of this section.

Figure 8. Schematic representation of facial musculature. Reprinted from Handbook of Psychophysiology (Cacioppo,

Bernston, & Tassinary, 2000, p. 706).

42


In terms of emotional expression, EMG is valuable as a measure of valence. Like EDA,

EMG is highly sensitive to changes in the CNS as a consequence of psychological stimuli; this is

evident in consistent patterning of specific facial muscle activation in emotional response through

facial expression (Andreassi, 2007, p. 300). Positive and negative valence can be determined

through the direct measurement of these specific muscles. One benefit of employing EMG to

measure facial expression is that it is capable of detecting changes in emotional processes that are

‘too subtle or fleeting to produce observable facial expression changes’ (Andreassi, 2007, p. 303).

Both the ZM and OO have been used to index positively valenced emotions such as joy

and excitement through the muscle

activation that occurs during smiling

(Andreassi, 2007, pp. 300-303; Ravaja,

Turpeinen, Saari, Puttonen, & Keltikangas-

Järvinen, 2008; Witvliet & Vrana, 1995). In

the ZM, this is through the lip corners being

pulled up and back; in the OO, it is through

skin tightening that causes the lower eyelid

to rise (Tassinary & Cacioppo, 2001, p.

166). Conversely, the corrugator supercilii

(CS) has been routinely correlated with

negatively valenced emotional expression,

such as anger and fear (Lang, Greenwald,

Bradly, & Hamm, 1993; Dimberg, 1990; Andreassi, 2007, pp. 300‒303) through the furrowing of

the brow (Tassinary and Cacioppo, 2000, p. 166). For the exact location of these muscles, see

Figure 8.

Figure 9. Suggested facial EMG electrode placement.

Reprinted from Handbook of Psychophysiology (Cacioppo,

Bernston, & Tassinary, 2000, p. 174).

43


As facial muscles are striated (fibrous) and exist within close proximity of one another,

the accurate determination of which muscles are contracting through cutaneous electrodes is

challenging. Tassinary and Cacioppo (2001, p. 167) recommend referring to the muscle sites as

regions; as opposed to ‘EMG activity in the CS’, it is more appropriate to refer to ‘EMG activity

to the CS muscle region’. Consistent application of EMG electrodes is necessary to ensure an

accurate signal from the correct muscle region (see Figure 9). As EMG activity is especially

susceptible to particle and movement interference, electrode preparation and attachment must be

undertaken carefully (Stern et al., 2001, pp. 112‒113).

2.8.7.3 Electrocardiography

Electrocardiography (ECG or EKG) is the measurement, through cutaneous electrodes, of

the electrical changes that occur during the heart’s contractions. Although an autonomic activity,

the frequencies, variations and pace of these contractions are responsive to the psychological

stimuli of stressors, frustrations and fears (Andreassi, 2007, p. 438‒439; Melillo, Bracale, &

Pecchia, 2011), and cognitive processing (Allen, Obrist, Sherwood, & Crowell, 1987; Szabo &

Gauvin, 1992). A primary interest in cardiovascular psychophysiology is stress as it relates to heart

disease.

44


The electrical changes of heart contractions present on a normal ECG as a composite of

P, Q, R, S and T waves, each representing a discrete minute physiological change during the

heartbeat (see Figure 10). The QRS complex is caused by ‘currents generated in the ventricles

during depolarization just prior to ventricular contraction’ (Andreassi, 2007, p. 412), with the R

wave its most prominent component. Due to its relative prominence, the R wave is the basis of

many ECG analysis methods.

Figure 10. A typical ECG trace and the associated physiological events. Reproduced

from Psychophysiological Recording (Stern, Ray, & Quigley, 2001, p. 181).

45


Heart Rate and Interbeat Interval

The HR, commonly measured in beats per minute (BPM), is often used in

psychophysiological research as a measure of increased arousal (Andreassi, 2007, p. 415). This is

based on the number of occurrences over the preferred time window of the aforementioned R

wave. An alternative method of approach is the interbeat interval (IBI), which measures the time

that has passed between each R wave. Typically, IBI is employed in the investigation of a single

cardiac cycle, whereas HR is preferred for analysis of over 30 seconds (Andreassi, 2007, p. 415).

HR increases have been found to occur during stressful events (Melillo, Bracale, &

Pecchia, 2011), the performance of cognitive tasks (Allen, Obrist, Sherwood, & Crowell, 1987;

Szabo & Gauvin, 1992) and during experiences of shock, fear and anger (Andreassi 2007, p. 440).

An intuitive example of a cardiac response to a psychological event may be that of a ‘racing heart’

in adverse or anxiety-inducing situations.

Heart Rate Variability

Heart rate variability (HRV) is a measure of autonomic activity that describes variations

in the HR through the measurement of the periods between beats over time (Task Force of the

European Society of Cardiology and the North American Society of Pacing and

Electrophysiology, 1996). As with HR analysis, the distance between R waves is measured;

however, of interest in HRV is the variation in the beat-to-beat (or R-R) intervals. HRV is typically

analysed in time or frequency domains (although other methods, such as geometric, do exist). The

time domain method may include investigating the difference between the longest and shortest R-

R intervals, the difference between HR in separate conditions and differences in spontaneous HR

response to stimuli (Task Force of the European Society of Cardiology and the North American

Society of Pacing and Electrophysiology, 1996). The frequency domain method identifies three

46


main spectral components of the HR: very low frequency (VLF), low frequency (LF) and high

frequency (HF). HF has been found to decrease under conditions that evoke time pressure,

emotional strain and state anxiety (Nickel & Nachreiner, 2003; Jönsson, 2007) and is considered

a direct measure of parasympathetic activation (Berntson et al., 1997); this is also true of HF peaks,

which are believed to reflect parasympathetic nerve activity (Billman, 2013). The role of LF/HF

ratio is contentious within psychophysiological literature, with research identifying it as an

inaccurate measure of cardiac sympatho-vagal balance (Billman, 2013).

Decreases in HRV indicate increased mental and physical stress (Schubert, Lambertz,

Nelesen, Bardwell, Choi & Dimsdale, 2009). For example, Melillo et al. (2011) compared the

HRV of university students taken during examination (a stressor) and immediately after returning

from university holidays (the non-stressor, or control session); a significant HRV decrease was

found in the stressor/examination session. Associations between decreases in the HF component

of HRV and increased mental workload and attentional focus have also been found (Cinaz et al.,

2013; Hjortskov, N., 2004).

47


2.8.7.4 Electroencephalography

EEG measures the electrical activity of the cerebral cortex as recorded from the scalp

(Stern et al., 2001, p. 79), allowing for insights into cortical activity that occurs during experiences

of attention, perception, sleep, sensation and emotional function (Andreassi, 2007, p. 115;

Davidson, Jackson, & Larson, 2000, p. 27). This electrical activity is generated by postsynaptic

potentials of cortical nerve cells that originate in the cerebral cortex (Sanei & Chambers, 2008).

EEG activity is processed via two parameters: amplitude (the size of the signal) and frequency

(how fast the signal cycles), allowing for the detection of patterns, or ‘bands’, that emerge in

cognitive activity (Stern et al., 2001, p. 80). The increases and decreases in power of the bands

represent differing neurobehavioural states, summarised in Table 1, are as follows:

Figure 11. EEG traces during various mental states. Reprinted from Psychophysiology:

Human Behavior & Physiological Response (Andreassi, 2007, p. 68).

48


Delta Activity (0.5–4 Hz)

Delta activity is a low-frequency band that only emerges in adults during deep, or slow

wave, sleep (see the lowermost trace ‘Deep Sleep’ in Figure 11). The analysis of this band is

generally exclusive to sleep research.

Theta Activity (4–8 Hz)

Theta activity has been found to occur in both states of drowsiness, rapid eye movement

(REM) sleep, problem-solving and attentional focus (Stern et al., 2001, p. 81). Presence of theta

activity has also been found in situations of inhibited response, in which a response to stimuli is

suppressed (Kirmizi-Alan, Bayraktaroglu, Gurvit, Keskin, Emre & Demiralp, 2006).

Alpha Activity (8–12 Hz)

Alpha activity occurs in awake, relaxing individuals, particularly when their eyes are

closed. The rhythms that emerge during this are called the ‘alpha rhythms’, and are associated

with relaxation and the relative lack of cognitive processes (Stern et al., 2001, p. 80). This state is

known as ‘relaxed wakefulness’ (Davidson et al., 2000, p. 31), and primarily originates in the

posterior and side regions of the head. Interrupting this state by, for example, asking an individual

to undertake a complex cognitive task or introducing a stressor, is known as ‘alpha blocking’

(Stern et al., p. 80). An alpha wave is demonstrated in the second wave from the top (‘Relaxed’)

in Figure 11.

Beta Activity (18–30 Hz)

Beta activity occurs during states of alertness, and is most common when an individual is

engaged in acts of mental or physical activity (Andreassi, 2007, p. 69). It is commonly associated

with states of active or anxious concentration (Baumeister, Barthel, Geiss & Weiss, 2008), and

indicates activation associated with cognitive task demands (Fernandez, Harmony, & Rodriguez,

49


1995; Ray & Cole, 1985). A typical beta frequency trace is found in the topmost trace (‘Excited’)

of Figure 11.

Gamma Activity (32+ Hz)

Gamma activity has been correlated with exposure to sensory stimuli such as flashing

lights or auditory clicking (Andreassi, 2007, p. 69), as well as the cognitive processing associated

with the distinction between non-important and important stimuli (for example, trying to find a

hidden object in a picture) (Stern et al., 2001, p. 81). However, some studies have suggested that

gamma activity is a by-product of electromyographic interference, such as ocular movement, and

not indicative of cognitive processing (Whitham et al., 2007; Whitham et al., 2008).

Table 1. EEG frequency bands and associated states.

EEG activity is obtained through both a multitude of single electrodes separately attached

to the scalp, or—more commonly—through a cap, headset or net system that contains all

electrodes (Davidson, Jackson, & Larson, 2000, p. 35). EEG headsets that feature as few as a

single electrode or channel are available, such as the NeuroSky MindWave headset (NeuroSky,

Band Frequency

)(

Associated States

Delta 0.5 – 4 Hz Deep sleep

Theta 4 – 8 Hz Drowsiness; REM; problem-solving; attentional

focus; inhibited response

Alpha 8 – 12 Hz Relaxed wakefulness; closed eyes

Beta 18-30 Hz Active or anxious concentration; task demands

Gamma 32+ Hz Exposure to sensory stimuli; cognitive processing

of visual stimuli

50


2015), but greater coverage is provided in 14-, 16- and 32-channel arrays, as well as high-density

256-channel systems (EMOTIV, 2016; Davidson, Jackson, & Larson, 2000, p. 35).

Electrode sites are chosen in accordance with the 10‒20 system (see Figure 12), which

allows for standardised placement of electrodes across the scalp. It is accepted research practice

to refer to the data collected by individual electrodes by their site in the 10‒20 system, which

identifies the region the electrical activity was collected from. The numbers on the 10‒20 system

indicate hemisphere (odd numbers = left, even numbers = right, z = midline), whereas the letters

indicate general cortical zone (O = occipital, P = parietal, C = central, T = temporal, F = frontal).

For example, activity obtained from the O2 site would refer to activity from the right hemisphere

of the occipital lobe. As both general cognitive processing and EEG bands originate in various

regions of the brain, the consistency of record-keeping in accordance with the 10‒20 system is

imperative for comparing findings across laboratories and studies (Davidson, Jackson, & Larson,

2000, pp. 32‒33; Andreassi, 2007, pp. 72–73).

51


2.8.8 Limitations

Psychophysiological assessment contributes several advantages to the process of evaluating

experiences and states: it allows for a covert, direct evaluation of emotional signals in response to

an event or experience (Nacke, 2013), facilitates real-time response analysis and provides

objective data that is unlikely to be influenced by intentional obfuscation or dishonesty. However,

there are limitations associated with the method—in particular, the complexities of collection,

analysis and interpretation.

Figure 12. International 10‒20 System. Reprinted from Psychophysiology: Human Behavior

& Physiological Response (Andreassi, 2007, p. 72).

52


The complex domain relationships of psychological states and physiological response

means that interpreting a physiological response as indicative of a specific psychological state is

often limited; this is further bolstered by the presence of directional fractionation.

Psychophysiological analysis—especially tonic analysis—thus benefits from the inclusion of

additional measures for comparison.

Set-up, data treatment and analysis of psychophysiological data also represent notable time

investments. Such investments for these processes within the current research program are detailed

in section 6.4.1. Financial costs signify an additional barrier to entry: initial purchase of clinical

psychophysiological equipment is a costly endeavour, with the resupply of relevant consumables

(such as electrode gel or disposable electrodes) creating additional ongoing costs throughout data

collection.

Finally, while psychophysiological assessment does allow for the covert recording of

physiological response (in that participants do not need to answer directly), placing electrodes on

surfaces of the body limits the naturalisation of the experiment. Furthermore, many instruments—

such as EEG caps—may be uncomfortable for the participant.

2.9 PSYCHOPHYSIOLOGY OF THE PLAYER EXPERIENCE

Since 2004, researchers such as Mandryk (2004, 2006a, 2006b, 2007), Nacke (2008, 2009,

2010a, 2010b, 2010c) and Ravaja (2006a, 2006b, 2008) have published multiple papers exploring

the relationship between psychophysiology and player experience. These papers have largely

successfully shown the value of psychophysiological measures as a means of monitoring player

experience.

53


Psychophysiology has also experienced growing popularity within the games industry.

Newell (2008), co-founder and managing director of the video game development company Valve

Corporation, has stated that ‘we’re really interested in … testing biometrics on player state’.

Ambinder (2011) reveals Valve’s adoption of psychophysiology not only in terms of evaluating

the player experience, but in creating games that respond dynamically to the player—such as a

dynamic artificial intelligence (AI) director for one of Valve’s flagship titles, Left 4 Dead 2 (2009).

Collaboration between industry and academic groups has also been present in video game

developer and publisher PopCap Games’ funding of Russoniello’s (2009a, 2009b)

psychophysiological studies of casual game effect on mood. The growing prominence of

psychophysiology within both research and industry necessitates an understanding of the more

popular psychophysiological measures’ capabilities and usage in the player experience context.

2.9.1 Validating the Psychophysiological Method in Games

A number of studies exploring the applicability of physiological measures and

methodologies in the player experience space, as well as the validation of physiological results in

the context of video games, have been undertaken; for a review, see Kivikangas et al. (2011).

These studies help situate the efficacy of psychophysiology within games research and evaluation.

Mandryk, Inkpen and Calvert (2006b), in the first of two experiments, explored whether

there were associations between a player’s physiological state, events occurring during the

entertainment experience and the subjective reported experience. EDA, ECG and EMG (ZM and

CS) were used in concert with interviews, questionnaires (on a custom Likert scale measuring for

challenge, frustration, boredom and fun) and video analysis. Mandryk et al. later discarded video

analysis due to the substantial time commitment required. Eight male university students, aged

20—26, played three versions of hockey game NHL 2003 (Electronic Arts, 2002): easy, medium

54


and difficult. No main effects of difficulty level were found on any of the physiological measures,

although some effects of player expertise were found: self-identified expert players had higher

respiration rates than the semi-experienced and novel players, self-identified novice players had

higher HR in the easy condition, and self-identified semi-experienced players had higher HR in

the difficult condition. The researchers later determined that the participants were responding

inconsistently to experimental manipulations (no differences in perceived difficulty were found

between the medium and difficult levels). Additionally, the researchers discovered that the process

of interviewing had lasting effects on psychophysiological response. In Figure 13, the lighter grey

columns represent the interviews; the effects of the interviews on arousal have affected the

preceding darker columns, which represent gameplay. The researchers suggest that various

interview issues (the gender of, or unfamiliarity with, the interviewer, personal space issues and

nerves) may have influenced the psychophysiological data.

In a following experiment, Mandryk and colleagues relegated the interviews to online

questionnaires so as to minimise human interaction effects on psychophysiological data. Vicente’s

(as cited in Mandryk et al., 2007) recommendation for the collection of baselines through the

experiment as a regulation for such events was also incorporated. As the primary aim of the study

was to explore the use of psychophysiology in player experience evaluation, the researchers also

introduced a more potentially effective manipulation to the experiment design by incorporating a

multiplayer game mode (play against a co-located player, and play against AI). ECG, HR and

EMG (ZM and CS) data was collected from 10 male participants aged 19—23. Tonic EDA was

significantly higher in the co-located play condition than in the AI condition, as was mean EMG

(the authors do not differentiate between sites). Physiological and subjective data was also found

to correspond: those who found the AI condition more challenging had higher tonic EMG in the

AI condition than in the co-located condition. EDA also proved useful for phasic analysis, showing

55


greater EDA in goals scored against the co-located player than those scored against AI. The

alterations to the experiment design thus proved helpful, and the researchers’ findings supported

the hypothesis that physiological results will correspond to subjective reported experience. It is

evident that a considered mixed-methods approach and a careful methodology can enable richer

insight into the player experience.

In a study designed to explore the utility of EEG for evaluating the player experience,

Nacke (2010) implemented EEG, the Game Experience Questionnaire (GEQ) and the MEC

Spatial Presence Questionnaire (SPQ) in an experiment investigating affective gameplay

interaction through a comparison of a PlayStation 2 (PS2) controller and a movement-based

WiiMote. This research was successful in discovering correlations between EEG and GEQ/SPQ,

suggesting that EEG is a valuable tool for understanding the neurology of the player experience.

Thirty-six university students (7 female) between the ages of 18—41 played first-person shooter

Figure 13. EDA during play and interviews, wherein the light grey columns represent interviews.

Reproduced from ‘Using psychophysiological techniques to measure user experience with

entertainment technologies’ by R. L. Mandryk, K. M. Inkpen, and T. W. Calvert, 2006, Behaviour

and Information Technology, 25(2) p.150.

56


horror game Resident Evil 4 (Capcom, 2005) in both a PS2 and WiiMote condition. Nacke found

a number of significant EEG band responses corresponding with interactions between participant

expertise and experiment condition. Additionally, gender was found to have moderating effects

on alpha, beta and gamma activity, due to a tendency for stronger psychophysiological response

in women (Cacioppo et al., 2000). Correlations were also found between EEG bands and

SPQ/GEQ, such as a positive correlation between alpha power and reported tension ratings.

In an investigation of measuring emotional valence in video game play, Hazlett (2006)

employed EMG ZM and EMG CS to measure psychophysiological responses to positive and

negative video game events. Thirteen male participants, aged 9—15 years old, played Xbox car

racing game Juiced (THQ, 2005) versus AI opponents. Hazlett found greater EMG ZM (the

‘smiling muscle’) activity in events identified in video review as positive (corroborated by

discussion with expert players), and greater EMG CS (the ‘frowning muscle’) activity in events

likewise identified as negative. These findings helped to situate EMG as an effective tool for the

measurement of positive and negative player experiences.

Finally, Ravaja, Saari, Salminen, Laarni and Kallinen (2006a) used only

psychophysiological (EMG, EDA and ECG) measures in conjunction with events in games to

determine phasic emotional reactions. Thirty-six university students (11 female) aged 20—30

years old participated in the study. Psychophysiological feedback was compared with events in

game, so as to confirm the hypotheses that unequivocally rewarding events—such as picking up

a banana within Sega’s Super Monkey Ball 2 (2002)—would elicit positively valenced arousal, as

indexed by increased ZM and OO EMG activity, increased EDA and increased HR. The

researchers were successful in proving these hypotheses, situating physiological measures as an

appropriate measure for player experience analysis. Ravaja and colleagues later employed

57


psychophysiological analysis in a study investigating phasic emotional responses to violent video

game events (Ravaja et al., 2008), as discussed in section 2.9.2.1.

These studies offer important implications for future research projects in the field of

psychophysiological evaluation. The success of the studies conducted by Mandryk et al. (2006b),

Nacke (2010b) and Ravaja et al. (2006a, 2008) point to the potential of a mixed-methods

approach—subjective and psychophysiological—in evaluating the game player experience. They

also help to identify methodological flaws that can erroneously influence psychophysiological

data, as in the case of participant‒researcher interaction.

2.9.2 Game Effects

2.9.2.1 Violence

A central focus of psychophysiological game evaluation is distinguishing the effects

different types of games or play experiences may have on players (Kivikangas et al., 2011). Of

particular interest is the effect of violent game content, both physiologically and psychologically.

Kivikangas et al. (2011) suggest that existing psychophysiological research in the space is

contradictory and often flawed methodologically, as a result of a tendency to compare subjective

and physiological responses to different games or different genres (e.g. one violent, one non-

violent). This lack of direct comparison introduces the risk of uncontrolled variables that influence

data.

One example of this is evident in studies undertaken by Fleming and Rickwood (2001)

and Ballard and West (1996), as reviewed in Kivikangas et al.’s (2011) meta-analysis of

psychophysiological games research. Fleming and Rickwood employed HR analysis, as well as

subjective measures of mood and arousal, to investigate the relationship between children’s mood

and violent video games. Seventy-one children (35 female) aged 8—12 participated in two non-

58


violent game conditions and one violent game condition. Both physiological (HR) and subjective

arousal were found to increase significantly after exposure to the violent game condition.

However, it is important to contextualise these findings: the two non-violent game conditions were

a pen-and-paper maze game and a geometric puzzle video game, and the violent game condition

was an action-adventure combat-based video game Herc’s Adventures (LucasArts, 1997), rated

G8+ (no parental guidance recommended for children aged eight or older). It is impossible to

determine whether the increased physiological response was to violent content, or other variables

that were not controlled for in the conditions—for example, graphical differences, story (present

in the violent game, but not the non-violent), audio differences and so forth. Furthermore, as

Herc’s Adventures features heavily stylised and cartoon-like graphics, the effect of the violence

may not be comparable with that of the violent games employed in other literature discussed here.

Similarly, Ballard and West (1996) compared HR and BP for 30 male undergraduates across

three conditions: one non-violent (billiards) and two violent (two versions of Mortal Kombat II

[Midway Games, 1993], one featuring less violent graphical content than the other). Again, higher

HR was found in the violent game conditions (Mortal Kombat II) than in the non-violent game

condition (billiards). The differences between billiards and video games are such that a direct

physiological comparison is problematic; as discussed in section 2.8.6.5, the many-to-one domain

relationship suggests that a single physiological response may be indicative of many psychological

events. More variables are controlled for in the direct comparison of the less and more graphically

violent Mortal Kombat II versions, in which participants experienced greater BP reactivity in the

‘more violent’ version.

Gentile, Bender & Anderson (2017) investigated the effects of violent content in video games

on physiological arousal, as measured by ECG and salivary cortisol, in 136 children (67 female)

59


aged 8—12 years old. As with Ballard and West (1996) and Fleming and Rickwood (2001),

Gentile et al. (2017) compared psychophysiological response to violent and non-violent games

Spiderman [Activision, 2002] and Finding Nemo [THQ, 2003]. Both games were rated as ‘equally

exciting’ by participants. Results revealed significantly increased cardiovascular arousal and

increased cortisol in the violent game than the non-violent game, and are suggested as indicative

of a fight-or-flight response in the participants. This methodology is strengthened by ensuring that

each game experience was considered ‘equally exciting’ by the participants, thus controlling

somewhat for gameplay variations as an explanation for the pattern of results; but, again, the

research is limited in the comparison of two notably dissimilar play experiences.

Kivikangas et al. (2011) highlight research undertaken by Weber, Behr, Tamborini,

Ritterfield and Mathiak (2009) as notable in addressing these variations through the direct

comparison of events—a phasic approach—within a single 50-minute first-person shooter video

game (Tactical Ops: Assault on Terror, Kamehan Studios, 1999) session across 13 experienced

male FPS players (aged 18—26) using EDA and HR measures. These events included situations

such as ‘Round begins’, ‘Runs out of ammo during combat’ and ‘Killed opponent’. While EDA

results did not reach significance, HR showed significant increases in response to certain events.

The events with the highest HR included ‘Player begins round’ and ‘Player ran out of ammunition

under combat’. The authors speculate that the former finding suggests participant uncertainty as

to their future performance and upcoming events, with the latter suggesting increased HR response

to imminent perceived danger. The former finding is particularly interesting for

psychophysiological methodologies within the video games space, as researchers may expect—

and should account for—arousal increase at the beginning of a game condition, regardless of the

intended game experience. Although the rigour of Weber et al.’s research benefits from the direct

evaluation of participant experiences within the same game condition, it should be noted that the

60


researchers allowed for the participants to freely select from 15 different maps. Weber et al. note

that the maps were comparable in size and player mission, and that the selection of maps allows

for a more naturalised player experience. Despite this, it is possible that differing map experiences

influenced psychophysiological response between participants.

In Ravaja et al.’s (2008) phasic analysis of violent video game events, EDA, EMG OO, CS

and EMG ZM activity was recorded from 36 (11 female) undergraduates, aged 20—30 years old,

in response to the wounding and death of the player-character and enemy characters within the

first-person shooter James Bond 007: Night Fire (Electronic Arts, 2002). In response to wounding

and killing an opponent, Ravaja et al. found increased EDA and decreased ZM and OO activity;

conversely, the wounding and death of the player-character evoked increased EDA, ZM and OO

activity, with a decrease in CS activity. These results suggest an increase in high-arousal negative

affect in response the killing and wounding of an opponent, and an increase in positive emotion

in response to player death. While this was primarily a study of trait psychoticism among

participants, with findings suggesting that those who rated highly on a psychoticism responded

less negatively to enemy wounding or death, these results also have interesting implications for

player experience—most notably that success in a game through defeating enemy units may be

associated with increased negative valence.

In a separate investigation of the effect of both violence and difficulty, Kneer, Elson and

Knapp (2016) employed four modifications of the first-person shooter game Team Fortress 2

(Valve Corporation, 2007): a low-difficulty, high-violence condition; a low-difficulty, low-

violence condition, a high-difficulty, high-violence condition; and a high-difficulty, low-violence

condition, all within the same map. A sample of 90 (69 female) university students with a mean

age of 24.47 was observed, with each participant playing one of the conditions. Kneer, Elson and

61


Knapp employed IBI and EDA to evaluate psychophysiological arousal activation in each of the

four conditions. The researchers did not find any significant effects on either IBI or EDA by

difficulty, violence or their interaction, and suggest corroboration with ‘previous research

indicating that violence in games does not substantially influence human behavior or experience’

(p. 142). The absence of differences in tonic psychophysiological response between these

conditions further strengthens the possibility of game or genre difference, rather than the

differences in violent content, as a primary contributor to disparate psychophysiological response

in previous research methodologies.

2.9.2.2 Stress

Video games have previously been employed extensively as stressor stimuli in

psychophysiological research. As such, video games have been found to consistently induce

altered HR response and BP and increased cortisol levels in research exploring violence, challenge

and stress (van der Vijgh, Beun, van Rood, & Werkhoven, 2015). Despite this, a meta-analysis by

Vijgh et al. (2015) of research employing video game stressor stimuli discovered specific and

consistent moderating functions of both game and study characteristics, accounting for 43% of

variance found in physiological response. Vijgh and colleagues state that ‘a digital game stressor

does not act as a stressor by virtue of being a game, but rather derives its stressor function from

its characteristics and the methodology in which it is used’ (p. 1080). The authors present this as

evidence for the case for standardisation within psychophysiological and games research.

Additional psychophysiological research has also found evidence of video games as a

form of catharsis or stress release. Russoniello, O’Brien and Parks (2009) employed EEG, HR and

HRV analysis, as well as subjective measures of mood, in an investigation of the effect of casual

games on mood and stress across 69 participants (no demographic information was provided by

62


the authors). Play of the casual video game Bejewelled II (PopCap Games, 2004) resulted in

decreases in left frontal EEG alpha, leading Russoniello et al. to conclude increased mood,

increased HRV and decreased HR, indicating decreased arousal contextualised by the subjective

mood measure as a decrease in stress. Additionally, the researchers support further multi-modal

use of psychophysiological and subjective measures.

2.9.3 Social Play

Psychophysiological evaluation of social play experiences, particularly in the comparison

of human and AI teammates and enemies, has been an additional area of investigation in player

experience evaluation. Mandryk and Inkpen (2004) represented one of the first of these efforts in

their employment of EDA, EMG ZM and HR in assessing co-located competitive play against a

friend and play against AI in the ice hockey video game NHL 2003 (EA Sports, 2002). For the

friend condition, results revealed significantly greater EDA and EMG ZM activity, aligning with

subjective experiences that revealed instances of increased fun, excitement and engagement. These

findings establish increased EDA and EMG ZM as reflecting social experiences in play, and

indicate a tendency for increased physiological activity in play against a friend than against a

computer. However, the study was limited in sample size (10 male participants aged 19—23);

additional exploration of these experiences is thus recommended.

Ravaja et al. (2006b) used ECG, EMG CS, OO and ZM across 99 (48 female)

undergraduates aged 19—34 to evaluate competitive play against a human friend, human stranger

and AI in the video games Super Monkey Ball Jr. (THQ, 2002) and Duke Nukem Advance (Torus

Games, 2002). Similar to Mandryk and Inkpen (2004), Ravaja et al. encountered higher levels of

psychophysiological arousal when playing against human-controlled players, present in the form

of shorter IBIs (or increased HR) and increased positively valenced emotional responses in the

63


form of increased OO and ZM activity. Greater anticipated threat was also reported in play against

humans, supported both by subjective measures and increased sympathetic arousal.

Finally, Lim and Reeves (2010) employed psychophysiological assessment to explore

competitive and cooperative play experiences with human or AI counterparts, employing EDA,

SCR (skin conductance response) and HR within a sample of 34 (17 female) university students.

Lim and Reeves discovered that players experienced greater EDA, SCR and HR activity when

playing with humans, irrespective of competitive or cooperative play. These results contribute to

extant psychophysiological literature on social play experiences, supporting research establishing

increased physiological arousal in play with humans than with AI (Mandryk & Inkpen, 2004;

Ravaja et al., 2006b).

An investigation into the psychophysiological effect of social play was undertaken

throughout the author’s Honours year, and expanded upon during the first quarter of the PhD

candidature (Johnson, Wyeth, Clark, & Watling, 2015). The psychophysiological effect of the

player experience was explored by comparing play sessions with AI teammates and play sessions

with co-located human teammates within the same map of the game Payday: The Heist (Overkill

Software, 2011). The study employed a mixed-methods approach to evaluation, using EEG in

conjunction with selected subscales from the Player Experience of Needs Satisfaction scale

(PENS) and GEQ questionnaires. Seventy-two (80.6% male) participants, between the ages of

18—31 and self-describing as having at least some experience with video games, undertook the

study.

In terms of brain activity, the study revealed that play with human teammates was

associated with greater activity in the beta, theta and alpha power bands than play with AI

teammates. The association of power bands with general mental states allowed for insight into the

64


psychophysiological experience of the players. As greater activity in the beta band is typically

associated with cognitive processes, decision-making, problem-solving and information

processing (Lehmann et al., 2012; Nacke et al., 2011), the study indicated that participants were

processing more information during play with humans. Similarly, theta has been linked to

engagement and challenge in video game play (Salminen & Ravaja, 2007), potentially revealing

that participants found play with human teammates more challenging or engaging than play with

AI teammates. Overall, the study suggests that play with human teammates involves greater

cognitive activity than play with AI teammates.

Including questionnaires in the study bolstered understanding of EEG associations with

subjective measures. It was found that greater activity in the beta band was associated with

increased experience of presence when playing with AI teammates, implying that an increase in

information processing is associated with feelings of presence. Associations were also discovered

between autonomy and greater beta and gamma band activity in the AI condition; this is potentially

attributable to increased freedom in decision-making and strategy. Johnson et al. (2015) propose

that these results only emerge in the AI condition, and as a consequence, increased ‘mentalising’

in the human teammate condition, wherein concern for human teammate’s motivations and goals

may have moderated the relationships between subjective and physiological results.

2.9.4 Immersion

While the concept of fun is difficult to quantify, concepts such as enjoyment, immersion,

engagement and flow (see section 2.9.6) provide a helpful lens through which to evaluate the

player experience. A primary concern of player experience research is to gain further insight and

understanding of these experiences. Nacke and Lindley (2008, p. 1) emphasise the importance of

investigating these concepts:

65


Not only do these terms currently lack well-accepted common meanings, but also for game

designers, clear and testable definitions of constructs such as immersion and flow would be

invaluable, since these are considered to be the holy grail of digital game design.

In pursuit of this objective, Nacke and Lindley (2008) employed subjective (a GEQ and

an SPQ) and psychophysiological measures—EDA and EMG (OO, ZM, CS)—in their evaluation

of immersion and flow in first-person shooters. Only the findings on immersion are reported in

this section (refer to the following section, 2.9.6, for flow). The researchers developed a boredom

condition, immersion condition and flow condition modified from first-person shooter Half-Life

2 (2004). Data was collected from 25 male participants aged 19—38 years old. While the flow

condition was successfully distinguished from the immersion and boredom conditions in both

psychophysiological and GEQ data, there was not enough evidence to discriminate between

experiences in the immersion and boredom conditions. Nacke and Lindley state that, while flow

and boredom are intuitively understood, immersion is less so; it may be that the condition design

was not able to accurately evoke a sense of immersion, or a distinct enough sense of immersion to

separate it from experiences of flow or boredom. The boredom condition generated greater

psychophysiological activity within EMG ZM, EMG OO and EDA; however, it is unclear whether

this suggests decreased physiological reactivity during immersion, or whether this result is an

artefact of the condition design. This establishes the difficulty of evaluating immersion, despite its

ubiquity within player experience research.

Nacke, Gimshaw and Lindley (2010a) undertook an additional assessment of the player

experience of first-person shooters through the psychophysiological and subjective evaluation of

the effect of game sound (‘sonic user experience’). Participants were 36 (7 female) undergraduate

students and university employees between the ages of 18—41. While a significant main effect of

sound was discovered in the subjective analysis (finding that the inclusion of game sound led to a

66


more positive player experience), no main or interaction effects of sound were found for either

EMG or EDA. Nacke et al. suggest that this may be because of the tonic approach used, in which

reactions to sound were ‘averaged’ out; the authors identify tonic evaluation as suitable for some

player experience analysis, but note that a phasic analysis may be more appropriate for evaluating

sonic experiences in digital games.

2.9.5 Dynamic Difficulty Adjustment and Biofeedback

Most single-player video games allow for the adjustment of difficulty across several static

options, typically in a spectrum ranging from ‘easy’ or ‘beginner’ through to ‘hard’ or ‘insane’;

however, this lack of flexibility may lead to a mismatch between player skill and the challenges

of the game (Huncke & Chapman, 2005). This issue is circumvented through the introduction of

DDA: reactive game systems that adapt dynamically to player performance, allowing for the

maintenance of challenge‒skill balance (Hunicke & Chapman, 2005). This adjustment may be

undertaken through such routes as the real-time manipulation of opposing AI units, quantities of

resources available to the player and map layout to either increase or decrease the game difficulty

based on player performance (Baldwin et al., 2013). The real-time nature of psychophysiological

recording has situated psychophysiological analysis as uniquely suited for employment within

DDA contexts (Baldwin, 2017). An extension of this application is also found in use of

biofeedback in video games, in which physiological response is used to directly and indirectly

control game interaction (Nacke et al., 2011).

Mirza-Babei et al. (2013) investigated the use of psychophysiology in the creation of

‘biometric storyboards’, in which events in a game were mapped to players’ physiological

responses and used to inform design decisions for further iterations of game development. Mirza

et al. employed EDA, EMG CS, ZM and OO activity in their assessment of the play experience;

67


once the data was collated, the researchers reviewed game video data with their 6 male participants

to identify positive and negative experiences. These results informed further iteration of the game

design. In later comparisons of both non-biometric and biometric storyboards, Mirza-Babei et al.

found that the use of biometric storyboards resulted in a greater number of requested design

iterations and increased confidence in these iterations. In an assessment of the final game designs

across 24 experienced male PC gamers aged 19—27 years old, results revealed that the gameplay

experiences informed by biometric storyboards were ‘significantly better’ as assessed by

subjective measures. Mirza-Babei et al. suggest that these results highlight potential benefits of

psychophysiological analysis in games user testing and development.

In a psychophysiological evaluation of adaptive game conditions in Tetris using both EDA

and HR, with a sample of 18 (4 female) university students aged 19—31, Wu and Lin (2011)

found that EDA signals were able to quickly and robustly adjust player stress changes. This was

revealed through increased EDA in response to challenging or frustrating tasks. Furthermore, Wu

and Lin suggest that the combination of EDA and HR may be used to identify a ‘negative stress’,

or ‘distress’, threshold, which may be useful for DDA games to indicate the reduction of

challenging or frustrating events. The authors add that the same interpretation may be made of

decreased EDA, which could indicate boredom and thus promote the inclusion of challenging or

frustrating events.

Nacke et al. investigated the use of ECG, EDA, EMG, respiration, temperature, and gaze

as a method for both indirect and direct control of player experiences and interactions (Nacke et

al., 2011). Ten participants (3 female) between the ages of 21—40 years old played two game

conditions, in which one game condition featured direct physiological control (e.g., increased run

speed through flexing leg muscle) and the other indirect physiological control (e.g., increasing

68


game speed in response to HR). Results revealed a preference for direct physiological control,

with authors suggesting the use of indirect physiological input as a ‘dramatic device’ for

alterations of the game world.

Finally, Negini et al. (2014) explored the use of affective states—as measured by EDA—

to control in-game difficulty, such that decreases in physiological arousal would prompt increased

game difficulty and vice versa. The researchers examined this across three conditions, in which

game difficulty was modulated through the real-time modification of the player-character, enemy

NPCS, or game environment; a control condition, sans affective adaption, was also included. 16

university students (15 male) between 18—32 years old participated in the study. Negini et al.

found that real-time adaption of game difficulty increases physiological player arousal, treated by

the authors as a measure of excitement. The authors also found reduced player enjoyment in the

condition in which enemy NPCs were adapted, and point to the moderation of the player-character

or environment as preferred routes for maintaining or ensuring positive play experiences.

2.9.6 Flow and Challenge

Despite the prominence of flow theory within the sociology and psychology fields, the

construct has rarely been examined from a psychophysiological perspective (Peifer, 2012).

Traditionally, flow has been evaluated with subjective measures such as questionnaires and

interviews after the experience of flow has ended. Peifer (2012, p. 139) states,

The important conflict here is that as soon as participants are asked for their experience,

they enter into self-reflection and leave the flow state … psychophysiology can provide

physiological flow indicators that are assessed during the activity without interrupting the

participant.

69


It is evident that the psychophysiological investigation of flow allows for a deeper

understanding of the flow experience, particularly as it grants real-time insight into the experience.

Like psychophysiological evaluation of flow outside the field of video games,

psychophysiological evaluation of flow experienced during gameplay is rare. As discussed in

section 2.9.4, Nacke and Lindley (2008) used EMG OO, CS and ZM, as well as EDA, gathered

from 25 male participants aged 19—38 years old, in their evaluation of flow within a first-person

shooter environment. The boredom condition was designed to induce boredom, largely by

ensuring that player skill outstripped the challenge of the level; this was achieved by providing a

large quantity of health and ammunition supplies, reducing opponent strength and diminishing the

number of enemies. By contrast, the flow condition encouraged the flow state through gradual

increase in game difficulty, interesting combat mechanics and ‘cool down’ spots to diminish the

risk of overwhelming players. Nacke and Lindley found a correlation between facial muscles

indicating positive valence and the flow condition; similarly, EDA response showed greater

activation in the flow condition than the boredom condition. It should be noted that Nacke and

Lindley’s boredom condition was intended to detract from immersion in the game; it achieved this

through repeating textures, dampened sounds a linear level. It is important to understand that this

study compared flow to boredom (assumed by Nacke and Lindley to be a combination challenge‒

skill imbalance and low immersion), as opposed to a pure comparison of flow and no/low flow.

A study of 32 male participants (aged 17—32 years old) by Kivikangas (2006) does not

support Nacke and Lindley’s (2008) psychophysiological findings, finding no relationship with

flow and facial muscles indicating positive valence. Kivikangas also detected no relationships

between physiological arousal, as measured by EDA, and flow. The author suggests that this may

be because of the length of the play sessions in their experiment design (approximately 40

70


minutes); there is a possibility that averaging over long periods of time may have reduced the

associations between flow and psychophysiological response due to the potential for habituation,

as discussed in section 2.8.6.1.

Keller, Bless, Blomann and Kleinbohl (2011) revealed that reduced HRV, indicative of

enhanced mental workload, was associated with challenge‒skill balance. Increased HRV was

associated with boredom assessed within the study, reflecting lower mental load. Similar to Nacke

and Lindley, Keller at al. created a boredom condition and a challenge‒skill balance condition in

which balance was dynamically maintained (referred to as a ‘fit’ condition). A third condition, in

which the challenges of the game overwhelmed the skill of the player, was also created to measure

stress. These conditions were tested across eight (four female) university students; further

exploration of challenge‒skill balance across a larger sample would allow for greater

interpretability of results. The game used in the experiment—a version of the game show Who

Wants to be a Millionaire—is arguably better categorised as a digital quiz. The relevance of this

study to the broader digital games literature may thus be limited. Despite this, Keller et al. suggest

that ‘flow experiences represent a distinct state that can be identified not only with self-report data

but also on physiological measures’ (2011, p. 852).

Keller et al. (2011) also employed the same experimental approach discussed in section

2.3.1 to evaluate salivary cortisol levels across three difficulty levels in Tetris, featuring a sample

of 61 male university students (Keller and Bless, 2008). Salivary cortisol is considered a robust

indicator of physiological response to stressful stimuli, since increased synthesis of cortisol occurs

in response to psychological and physical stress (Kirschbaum & Hellhammer, 2000). Results

revealed relatively high levels of salivary cortisol in both the fit and overload conditions, with no

significant differences revealed between either; Keller et al. thus propose that flow experiences

71


may indicate increase physiological stress. They suggest that this may be a form of ‘positive

stress’; however, an elevated mood state was not corroborated by subjective measures employed

within the study.

Harmat et al. (2015) undertook a psychophysiological assessment of flow using similar

game conditions developed in Tetris, as in research by Keller et al. (2008; 2011). Featuring a

sample of 77 (40 female) participants (mean age of 27.8), Harmat et al. employed functional near-

infrared spectroscopy (fNIRS; a measure of oxygenation in the prefrontal cortex), ECG (HR and

HRV) and respiratory analysis to assess the physiological correlates of the flow experience across

easy, optimal and difficult conditions modulated by challenge‒skill balance and imbalance.

Results revealed higher flow experiences were associated with larger respiratory depth, which

Harmat et al. suggest indicates an association between increased flow and more relaxed

physiological states; furthermore, the authors propose an association of reduced LF and increased

flow to suggest deeper immersion in a flow-like state, as supported by previous research finding

decreased LF in meditation (Krygier et al., 2013). No relationships were discovered between

prefrontal cortex oxygenation, HR or the HF component of HRV.

Drachen, Nacke, Yannakakis and Pedersen (2010) used HR and EDA in their assessment

of physiological correlates of player experiences from 16 participants—no demographic

information was provided by the authors—in three first-person shooter games. In their approach,

Drachen et al. explored several psychological constructs of play, including flow and challenge,

within a single subjective scale. HR was found to correlate negatively with flow and challenge,

with a low HR indicating increased flow and challenge experiences; however, no significant

correlation was found between EDA and flow, or EDA and challenge. Drachen et al. propose the

72


absence of results for EDA and flow/challenge as a function of imprecise wording for the scale

used; this emphasises the difficulties of determining challenge in player experience evaluation.

2.9.7 Gaps Identified in Player Experience Literature

Overall, the physiological response to flow and challenge within video games is not yet

well understood. While some literature in the field generally points to physiological response,

indicating mental stress as associated with flow and challenge‒skill balance (Drachen et al., 2010;

Keller et al., 2011), results from other research does not support this (Kivikangas, 2006).

Understanding the role of positive valence in the assessment of flow and challenge is also limited,

with contradictory results emerging from player experience literature (Kivikangas, 2006; Nacke

& Lindley, 2008). Additionally, with occasional exceptions, psychophysiological assessment of

the player experience is often limited by small sample sizes and a lack of uniformity across

methodologies and experiment designs (see Table 2 for a summary). As such, while

psychophysiological measurement represents a keen area of interest for player experience

research, both the employment of psychophysiological measures and the results reported are often

inconsistent and somewhat limited by methodology and sample size. This points to an obvious

need in psychophysiological player experience literature for cohesive research that addresses these

limitations through implementation: one that aligns with recommended psychophysiological

practice, employs a large sample size, and includes the use of most commonly used

psychophysiological measures to investigate a construct or constructs of the player experience.

A unified psychophysiological analysis of the concept of flow and challenge would

benefit games literature by allowing for a richer understanding of flow (enabling insight into the

moment of flow), which would improve current understanding of the player experience. A

73


successful psychophysiological analysis would also introduce psychophysiology as a viable and

proven method for evaluating flow in future research and play-testing settings.

Table 2. Overview of research discussed in Chapter 2

Authors Year Measures N Area of Investigation

Ballard & West 1996 ECG (HR), BP 30 (M) Violent v. non-violent games,

hostility

Drachen et al. 2010 HR, EDA 16 Flow, challenge, enjoyment,

tension

Fleming &

Rickwood 2001 ECG (HR) 71 Violent v. non-violent games,

aggression

Gentile et al. 2017 ECG (HR, arterial

pressure), Cortisol 136 Violent v. non-violent games

Harmat et al. 2015 fNIRS, ECG, resp. 77 Flow

Hazlett 2006 EMG (CS, ZM) 13 (M) Valence during events

Johnson et al. 2015 EEG 72 Play with AI and humans Keller & Bless 2011 Salivary cortisol 61 (M) Flow

Keller et al. 2008 ECG (HRV) 8 Challenge‒skill balance Kivikangas 2006 EMG, EDA 32 (M) Flow

Kneer et al. 2015 EDA, ECG (IBI) 90 Violence, difficulty, affect

Lim & Reeves 2010 EDA, SCR, ECG

(HR) 34 Competition and cooperation

with humans and AI

Mandryk et al. 2006 EDA, ECG, EMG 8 (M) Difficulty

Mandryk et al. 2004 EDA, ECG, EMG 10 (M) Difficulty, play with AI and

humans

Mirza-Babei et al. 2013 EDA, EMG 6 (M) Biometric storyboards

Nacke & Lindley 2008 EMG, EDA 25 (M) Immersion, flow, boredom

Nacke 2010 EEG 36 Controllers

Nacke et al. 2010 EDA, EMG 36 Sonic experience

Nacke et al. 2011 EDA, ECG, EMG,

resp., temp., gaze 10 Direct and indirect

physiological control

Negini et al. 2014 EDA 16 Physiologically-adapted DDA

Ravaja et al. 2006 EMG, EDA, ECG 36 Phasic responses to video

game events

Ravaja et al. 2008 EMG, EDA 36 Phasic responses to violent

video game events

Russoniello et al. 2009 EEG, ECG (HRV) 69 Stress, mood

Weber et al. 2009 EDA, ECG (HR) 13 (M) Phasic responses in violent

games

Wu & Lin 2011 EDA, HR 18 Assessment of DDA

M = male participants only

74


3 RESEARCH DESIGN AND METHODOLOGIES

3.1 RESEARCH STRUCTURE AND SCOPE

As explored in the literature review, psychophysiological evaluation offers rich insight into

the player experience. However, the complexities and considerable temporal costs associated with

the psychophysiological method have directly influenced its employment within games research

and literature; despite the advantages of this approach, psychophysiological games studies have

typically featured small sample sizes, or restricted psychophysiological investigation to few

psychophysiological measures (see section 2.9.7). Disparities in methodologies have also been

identified as a potential source for discrepancies in findings, creating demand for a single unified

approach to the psychophysiological analysis of the play experience (Bernhaupt et al., 2008).

To this end, a rigorous program of research exploring the psychophysiological effect of play

was undertaken. The aim of this research is driven by gaps observed within existing

psychophysiological player experience literature, with the intention of strengthening and

contributing to the contemporary understanding of the psychophysiological experience of play.

Research Aim

1. To further clarify existing contributions to literature by expanding understanding of the

psychophysiological experience of play, and the value of psychophysiological

measures as a means of assessing the player experience.

To achieve this aim, the stringent application of psychophysiological method—as informed

by practice in the larger (beyond player experience) psychophysiological research space—was

essential. In concert with the existing psychophysiological player experience literature, the most

75


commonly used measures are deployed in this program of research. Furthermore, to ensure

relevance to player experience research, an approach employing both subjective and objective

measures was undertaken; this allows for an informed physiological exploration of self-reported

psychological constructs (as assessed by validated scales) already investigated in the context of

player experience. These constructs include phenomena such as flow, affect and needs satisfaction,

established in Chapter 3.

A set of criteria for the study methodology was established to enable a robust exploration

of the research aim and address gaps identified in the literature. The final methodology was

informed by the following considerations:

a) feature a large sample size composed of males and females

b) employ multiple measures of physiological assessment that may be feasibly deployed

in games research and play-testing

c) consider previous research in both psychophysiological and play experience literature,

and develop a methodology congruent with practice in the psychophysiology field

d) employ a video game artefact that represents the contemporary standard of commercial

games in order to maximise generalisability

e) explore prominent psychological constructs of video game play, as identified in the

literature.

The approach chosen was a large-scale single study, featuring both psychophysiological

and self-report measures and informed by a smaller scale pilot study to ensure successful

experiment and condition design. The final study would also be closely iterated throughout an

extensive design stage, as shaped by the pilot study aim, methodological considerations and

outcomes. This also allowed for the psychophysiological evaluation of multiple subjective player

76


experience constructs within the same framework, ensuring a consistent experience as necessitated

by the complex domain relationships in psychophysiological analysis (see section 2.8.6.5). Of

additional importance was the evaluation of a large sample size, established as a gap within the

existing literature.

A distinct element of this research program was the study format. In lieu of two to three

separate studies with smaller sample sizes of approximately 30–40, commensurate with existing

literature, the decision was made to employ a single large-scale study featuring a greater sample

size (approximately 90) and both psychophysiological and subjective measures. This would both

directly address the dearth of large-scale psychophysiological studies in player experience, and

allow for greater statistical power in analysis.

A key risk of this large-scale approach is the potential for incongruent experiment design

to impede, eliminate or otherwise influence both the psychophysiological and subjective

responses. This risk is minimised by an iterative design process throughout the development of

both experiment methodology and software. An additional hazard lay in the practicality of the

large sample size; previously, studies have been necessarily limited by the time costs associated

with psychophysiological deployment and data analysis. This is best combatted by including a

robust recruitment process and allotting adequate time to both data collection and analysis. The

strengths and limitations of the methodology, as well as the processes and iterations undertaken

for improving or ensuring design suitability and appropriateness, are further detailed in sections

6.4–6.5.

The expectation is that this research will expand current understanding of the

psychophysiology of video game play, and address gaps identified within existing literature in the

space. An additional expected contribution of this work is to provide insight into the logistics

77


associated with contemporaneously employing multiple psychophysiological measures in player

experience analysis, with direct implications for using psychophysiology in player experience

research and commercial settings.

3.2 RESEARCH STAGES

To satisfy the aim detailed in section 3.1, and further contribute novel discussion to both

games and psychophysiological literature, a multi-faceted research question was first established

to direct the design of a single large-scale investigative study:

Research Questions

1. How effectively can psychophysiological measures be used to evaluate the player

experience?

a. What are the differences in psychophysiological response between optimal and

sub-optimal play experiences?

b. Which psychophysiological measures, or combination of psychophysiological

measures, most reliably predict specific components of the player experience

as assessed by subjective measures?

Guided by these questions, the program of research adopted a four-stage process of design,

iteration, data collection and analysis (see Figure 14). To satisfy the requirements of RQ1a, both

optimal and sub-optimal play experiences were developed as video game artefacts within stages 1

and 2.

78


3.2.1 Summary Stage 1—Development of Methodologies and Game Artefact

As one aim of the research program was to expand upon the psychophysiological

understanding of the player experience and its application to evaluation, it was imperative to

identify both the prominent psychological constructs for investigation and the most effective

approach for the psychophysiological examination of these constructs. Examining play experience

literature revealed flow as a source of considerable interest within games research, applicable to

psychophysiological investigation and evaluable within the study scope.

STAGE 1

•determine a distinct study design for the comparison of physiological responses

•design and develop a video game artefact representative of a contemporary commercial game

•identify a viable physiological approach

STAGE 2

•pilot game artefact to determine effectivity; iterate and refine upon the design

•develop an experimental design congruent with psychophysiological measurement

•develop software that allows for partial automation

STAGE 3•data collection for final study

STAGE 4•investigation of RQs 1a and 1b

Figure 14. Research stages.

79


To investigate the psychophysiological experience of this construct, it was determined

that two conditions capable of (respectively) inducing and inhibiting these experiences would be

beneficial. This would allow for the direct comparison of physiological response, facilitating a

refined identification of the response source, or lack thereof. In the early stages of this program of

research, a focus on flow (to the exclusion of other subjective player experience constructs) was

considered, but subsequently identified as problematic during the piloting stage (see section 4.3).

It was also integral that the video game artefact chosen should represent contemporary

commercial games and thus be generalisable across the existing play experience landscape. To

this end, Valve Corporation’s popular first-person shooter Left 4 Dead 2 (2009) was selected and

modified for experimentation to have two conditions: a sub-optimal condition in which skills

significantly outstripped demands (the Boredom condition, expected to lead to low levels of flow),

and one in which the game dynamically ensured challenge‒skill balance (the Balance condition,

expected to lead to higher levels of flow).

Finally, a viable psychophysiological approach was identified based on deployability,

temporal efficiency, ease of access and interpretation, and potential for adoption within play

experience evaluation. The chosen approach consisted of EEG, EDA, ECG, EMG (ZM), EMG

(CS) and respiration; respiratory analysis was later removed from the methodology in Stage 2 due

to the respiratory belt’s tendency to overlap the ECG electrode situated on the rib.

3.2.2 Summary Stage 2—Pilot Testing and Design Iteration

A third sub-optimal condition, in which the demands of the condition outstripped the skill

of the player, was introduced to represent the spectrum of challenge‒skill balance and imbalance

(the Overload condition). The addition of this condition also represented a more nuanced approach

to exploring optimal and sub-optimal play experience, as necessitated by RQ1a.

80


The penultimate iteration of the video game artefact was pilot tested in an approximation

of the final methodology, without psychophysiological measurement, to determine the

effectiveness and reliability of the condition design. The pilot test surveyed participants on their

experience of flow during the three play conditions, revealing, unexpectedly, no significant

differences in flow between the Boredom and Balance conditions (see Chapter 4 for details).

These results informed the final iteration of the video game artefact design; possible

explanations for these findings suggest either an error in the development of the game conditions,

or possible limitations of the flow scale for the purposes of evaluating flow in a video game with

low levels of challenge. Potentially engaging elements of the Boredom condition were thus

removed to ensure a sub-optimal play experience. Furthermore, the scope of the research program

was broadened to include evaluation of presence, autonomy, competence, enjoyment and affect as

well as flow (in a reduced capacity); again, this allowed for a fuller definition of optimal and sub-

optimal player experience, strengthening exploration of the aim and research questions. Some

minor adjustments were made to the Balance condition to ensure a similar experience among all

participants. The experiment procedure was also refined for final deployment.

Finally, sequencing software was developed to minimise interaction with participants,

reduce experiment runtime, ensure accurate time lengths for all conditions and baselines, and

allow a single researcher to facilitate the experiment. This software was deployed for data

collection after extensive testing.

3.2.3 Summary Stage 3—Data Collection and Analysis

The final iteration of the study, including psychophysiological measures and the

additional subjective measures, was deployed alongside the sequencing software. The final

methodology is detailed in section 3.5 and Chapter 5. Data collection took place over a period of

81


13 months (March 2015 to April 2016), resulting in a sample size of 90 male and females aged

17+. Data was intermittently cleaned and evaluated throughout collection to ensure the quality of

the physiological data, identify potential trends and report preliminary results.

3.2.4 Summary Stage 4—Evaluation

Data was prepared and cleaned in accordance with the suggested methods for

physiological data treatment, with movement and noise artefacts—as identified by both visual

scanning and thresholds set within the analysis software—removed prior to analysis (for a full

write-up of the time costs associated with this process, see section 6.4.1). A spectral analysis of

two EEG sites, tonic mean analysis of EDA and EMG, and HR and HRV analysis of ECG was

performed. Analysis included a multivariate analysis of variance (MANOVA), analysis of

variance (ANOVA), correlations and regressions. This stage featured the investigation of RQ1a

and 1b, and the research aim.

82


3.3 STAGE 1—DEVELOPMENT OF METHODOLOGY AND ARTEFACT

Identification of a study design; selection of physiological approach; video game artefact design

and development

3.3.1 Introduction

As discussed in section 3.1, the aim and research questions of this research program

stipulate expanding upon existing research by evaluating psychophysiology in three contexts:

evaluating psychophysiology as a predictor of the subjective experience; the differences in

psychophysiological response between optimal and sub-optimal play experiences; and exploring

the value of psychophysiological measures for assessing the player experience. RQ1a further seeks

to explore what differences may emerge in the psychophysiological response to both optimal and

sub-optimal play experiences. It was also established that the research should achieve this by

identifying and resolving pertinent gaps in psychophysiological play experience literature.

To this end, it was established that the program of research should feature a large sample

size, multiple physiological measures, relevant subjective measure(s) and a contemporary video

game artefact, and be rooted in established psychophysiological method. The objectives of Stage

1 are as follows:

determine a distinct design for the comparison of physiological responses (optimal v. sub-

optimal)

select relevant subjective constructs

design and develop a video game artefact representative of a contemporary commercial

game

identify a viable physiological methodology.

83


3.3.2 Theoretical Grounding for Study Design

Due to the nature of domain relationships in psychophysiological measurement,

wherein—for example—one physiological response may indicate multiple psychological

processes (see section 2.8.6.5), a comparison of physiological responses to differing stimuli was

undertaken for the purposes of isolating potential relationships. Of priority to the program of

research (as identified in the research questions) is an exploration of how physiological response

may differ between exposure to optimal and sub-optimal play experiences, and whether

physiological response may be used to predict subjective experience of play experience

phenomena. It was intended that this would aid in providing a metric by which to evaluate the

experience of video game play through the patterns identified in physiological responses.

To more accurately isolate these relationships from multiple influences, it was determined

that the video game conditions should differ in a quantifiable way. Csikszentmihalyi’s (1990)

theory of the optimal experience through flow was thus adapted for the program of research.

Despite the prominence of the flow theory within player experience research and evaluation (Cox,

Cairnes, Shah, & Caroll, 2012), extant psychophysiological studies of flow in video games have

so far employed relatively limited sample sizes of 25 and eight (Nacke & Lindley, 2008; Keller &

Bless, 2008). A program of research that collated data from a larger sample size would not only

have the potential benefit of exploring and substantiating the conclusions of previous research, but

also contributing its own findings of additional relationships discovered through improved

statistical power. Employing a greater number of physiological measures, some potentially novel,

than those used in previous research would also provide the opportunity to expand current

understanding of the psychophysiology of flow in the play experience.

84


As in previous research, the current program of research employed multiple artefact

conditions for the purposes of psychophysiological comparison and prediction. Nacke and Lindley

(2008) used three game conditions to induce immersion, flow and boredom in a first-person

shooter video game; of relevance to this research are the modifications made to the flow and

boredom conditions. In Nacke and Lindley’s research, flow was designed for by including

challenges focused on ‘interesting combat mechanics’, gradual increase in difficulty and

‘cooldown spots’ (areas that allow players to rest and restock with sparse ammunition and health

supplies). The boredom condition focused on reduced challenge (weaker enemies and high

amounts of health and ammunition supplies) and a ‘boring’ play environment (repeating textures,

linearity and a limited choice of weapons). Nacke and Lindley’s boredom condition was rated by

participants as low on challenge, immersion and flow. Conversely, the flow condition scored

highest on flow, challenge and tension. The flow condition also elicited greater physiological

arousal and greater positive valence than the boredom condition.

Keller and Bless (2008) achieved flow and low-flow conditions through direct

manipulation of challenge‒skill balance in a Tetris-type video game. Flow was designed for in a

condition referred to as ‘adaptive’, in which task demands automatically adapted to player skill

through dynamic response to performance metrics. Low-flow states were achieved via boredom

and overload conditions. In the boredom condition, blocks fell at a very slow speed regardless of

player ability. In the overload condition, blocks fell at a fast base speed, which would continually

increase during play. Keller and Bless’ manipulation was successful in achieving greater

experience of flow than the low-flow conditions, as indicated by participant reports of several core

flow characteristics. Participants playing the adaptive condition experienced an altered perception

of time, perceiving the adaptive condition to be shorter than the boredom and overload conditions.

Participants also reported greater enjoyment and involvement in the adaptive condition than in the

85


boredom and overload conditions. This study identified challenge‒skill balance as a significant

predictor of intrinsic motivation.

The condition designs undertaken in these studies informed the condition design within

the current research program, representing a refinement and iteration of methodologies towards

the psychophysiological evaluation of video game flow.

3.3.3 Identifying a Viable Physiological Approach

The selection of physiological measures for evaluation was predominately motivated by

considering the availability and suitability of such measures in player experience research and

application domains, such as commercial studio play-testing. It was also anticipated that

identifying a robust, temporal and financially effective, and deployable psychophysiological

method may encourage further adoption of psychophysiological evaluation within player

experience research contexts. Determining the physiological measures to be employed was

influenced by the following criteria:

Accessibility: the measurement should be accessible, or potentially made accessible, to

developers or researchers not otherwise familiar with psychophysiological evaluation; it

should not be restricted to clinical or hospital environments, such as function magnetic

resonance imaging (fMRI).

Ease of deployment: researchers and developers not previously familiar with the

measurement should be able to be trained in its deployment.

Rate of employment within existing psychophysiological play experience literature: the

measurements chosen should mirror those typically used in extant literature.

Applicability of the measure to the evaluation of the video game experience: it should

not restrict necessary movement, such as hand movement, or impede visual senses;

86


similarly, it should be deployable within a computer or console lab environment (see

‘Accessibility’), and not overly obtrusive to the participant (as in the case of subcutaneous,

or needle, electrodes).

Complementary: the instruments themselves should be compatible with one another—

for example, they should not physically overlap one another.

Associated financial and temporal costs: the financial and temporal costs (such as set-

up) should be reasonable for general adoption within both research and industry.

EEG, ECG, EMG OO, EMG CS and EDA were chosen for their potential contributions

to expanding current understanding of the psychophysiology of the optimal play experience, as

well as for their adherence to the criteria established for selecting the psychophysiological

approach. With the exception of EEG, all measures were non-mobile and required a wired

connection to amplifiers. Some potential measures considered for inclusion were ultimately

discarded due to incompatibility with other measures, or inappropriateness or infeasibility within

wider research and industry contexts. These measures included EMG ZM, respiration and salivary

cortisol. In the case of EMG ZM, the location of the ZM muscle was determined to be infeasible

due to the inability to adequately prepare the site (e.g., shave) for analysis; measurement of

respiration through a respiratory belt proved to overlap with ECG electrodes, introducing a

potential for the dislodgement of electrodes or movement artefacts; and salivary cortisol was

deemed impractical for commercial research contexts due to hygiene concerns. Finally, other

measures—such as fMRI and functional near-infrared spectroscopy (fNIRS)—were not

considered due to expense or incompatibility with video game play.

A tonic approach was chosen to allow for assessment of the overall player experience.

While the use of phasic analysis has facilitated the evaluation of physiological responses in player

87


experiences (Ravaja et al., 2008) and been recommended for continued assessment of response to

specific in-game stimuli (Nacke et al., 2010a), the average tonic overview of the player experience

has been identified as capable of yielding suitable results for the analysis of player experiences

(Nacke et al., 2010a). The tonic approach has also been used regularly in previous player

experience research (Drachen et al., 2010; Harmat et al., 2015; Kivikangas, 2006; Kneer et al.,

2016; Mandryk et al., 2006b; Nacke & Lindley, 2009; Nacke et al., 2010a; Nacke et al., 2010b;

Russoniello et al., 2009) for both the insights it provides for the overall understanding of the player

experience and the reduced time costs relative to a phasic approach (Mandryk et al., 2006b).

Within psychophysiology, phasic assessment of responses to specific stimuli is a favoured

approach; however, Stern et al. (2001, p. 50) warn that ‘this emphasis on stimulus-contingent

responses and relative neglect of tonic levels is analogous to not seeing the forest for the trees’.

Due to these time costs, in association with the large sample size, a phasic analysis was determined

as beyond the scope of this research (although it nonetheless presents an opportunity for

exploration in future research).

3.3.3.1 Electroencephalography

The potential contributions of EEG to an understanding of the optimal play experience

are extensive due to the insights the measurement allows into cortical activity. Of particular

interest to player experience evaluation are the possible relationships that may be identified

between the play experience and EEG frequency bands, allowing for the exploration of associated

mental states such as attentional focus, information processing, mental workload and emotional

function in video game play.

A dry cap EEG headset was chosen due to its relative ease of deployment, applicability to

a video games research environment and unobtrusive nature. While wet cap EEG arrays allow for

88


direct contact with the scalp and are generally used in clinical settings, the set-up procedure is

involved, often time-consuming and can cause discomfort to the participant due to residual

electrode gel being left in their hair. The EEG device chosen for this program of research was the

EMOTIV Epoc+, a 14-channel wireless EEG that employs electrodes—soaked with saline

solution prior to use—with reusable electrode pads (see Figure X). The use of saline solution

instead of electrode gel or paste is helpful in reducing participant discomfort. The EMOTIV Epoc+

is paired with recording software ‘EMOTIV TestBench’, allowing for the real-time recording of

the raw EEG signal and the insertion of markers to represent events (such as the initiation or

termination of a condition). The EEG signal is sequentially sampled at a rate of 128 samples per

second (SPS).

While the set-up period of the headset was found to differ between participants due to

variables such as hair volume, initial testing determined that set-up would typically not exceed

five minutes in length. This situated the EMOTIV Epoc+ as a time-effective psychophysiological

measurement during deployment, despite the temporal costs associated with data treatment and

analysis (detailed in section 6.4.1). While not necessary for this program of research, the wireless

nature of the EMOTIV Epoc+ headset allows for its deployment in more mobile research

contexts—for example, the evaluation of exercise games that may require greater range of

movement. The headset was also chosen for its generalisability to wider research and play-testing

contexts.

A frequency analysis of the EEG data was chosen due to the interpretability of the

frequency bands in terms of player experience. The absolute power of the alpha, beta and theta

frequency bands were assessed for the associated cognitive states’ applicability to the assessment

of optimal and sub-optimal play conditions.

89


The association of alpha activity with ‘relaxed wakefulness’ (Davidson et al., 2000) and

relative lack of cognitive processes (Stern et al., 2001) offers the potential to investigate alpha

differences between a boring, or non-stimulating, play experience and stimulating ones;

furthermore, decreases in alpha activity have been previously identified in players as indicating

increased mood (Russoniello et al., 2009.) It was expected that beta activity, associated with

alertness, mental activity and responses to cognitive task demands (Andreassi, 2007, p. 69;

Fernandez, Harmony, & Rodriguez, 1995; Ray & Cole, 1985), would also reveal interesting

differences between optimal and sub-optimal play experiences. Finally, as increased theta activity

has been previously associated with increased challenge in play experiences (Salminen & Ravaja,

2007), this represented a natural path for further analysis in the context of manipulated challenge‒

skill balance; furthermore, as theta activity has been simultaneously associated with states of

drowsiness and attentional focus (Stern et al., 2001, p. 81), additional investigation of this band in

a player experience context may expand upon psychophysiological understanding.

As delta activity only emerged in adults during slow wave sleep, this band was excluded

from analysis (Stern et al., 2001, p. 81). As some studies have suggested that gamma activity may

be a by-product of electromyographic interference (such as ocular movement), rather than

indicative of cognitive activity, this band was also excluded from analysis (Whitham et al., 2007;

Whitham et al., 2008); however, future research may benefit from exploring this frequency band

in the context of player experience analysis.

Finally, the sites chosen for analysis were the AF4 (frontal) and O2 (occipital; rear) right

hemispheric sites (see Figure 12, p. 50). The AF4 site was chosen as beta activity—associated

with mental processing and alertness, above—has been identified as most evident in the frontal

cortex (Nacke, 2010), and has consequently been the focus of evaluating cognitive activity in

90


player experience research (Nacke, 2010; Johnson et al., 2015; Russoniello et al., 2009). The O2,

a site located on the occipital lobe, was chosen due to the association with increased occipital theta

and challenge in a previous psychophysiological evaluation of the player experience (Salminen &

Ravaja, 2008). In the interest of time constraints, no other sites were included in analysis; however,

these represent a path for analysis in future research. The reference channels for the EMOTIV

Epoc are in the P3 and P4 locations.

3.3.3.2 Electrodermal Activity

As EDA is widely regarded as one of the most easily deployable and interpretable

measures in psychophysiological analysis (Stern et al., 2001), this measure was chosen due to its

potential for generalisability to both research and industry environments. EDA was also chosen

due to its demonstrable robustness as a measurement of physiological arousal. The prevalent

employment of EDA in video game biofeedback research also establishes it as a measure relevant

to player experience evaluation contexts.

The EDA system employed for this program of research was supplied by BIOPAC

Systems Inc., and consisted of an EDA amplifier that uses a constant voltage of 0.5 V to measure

skin conductance, disposable snap 27 mm x 36 mm Ag/AgCl electrodes with a pre-filled gel cavity

and unshielded electrode leads 1 m in length. The unshielded leads were chosen due to the use of

an EMG ground electrode positioned directly in the centre of the forehead, as recommended by

Fridlund and Cacioppo (1986). EDA data is recorded and analysed in BIOPAC’s AcqKnowledge,

software paired with all BIOPAC systems; as in EMOTIV’s TestBench software, AcqKnowledge

records raw psychophysiological data in real time and allows for the insertion of markers to

indicate events.

91


The sites chosen for EDA analysis were the hypothenar and the thenar eminences of the

palm. While the distal phalanges feature the greatest concentration of eccrine sweat glands, as

established in section 2.8.7.1., this site was disqualified in this study due to the potential for

excessive movement artefacts associated with rapid and continual finger movement during

computer gaming.

Set-up time for the EDA system was not found to exceed five minutes, partially due to the

minimal preparation required for the EDA electrodes—unlike with EMG and ECG, no abrasion

is necessary; participants are simply required to wash their hands. The pre-filled disposable

electrodes also aided in the brevity of the set-up period. Despite the compact set-up period, a

further five minutes is required to elapse prior to recording to allow for successful skin‒gel contact

to be established (Braithwaite et al., 2015).

3.3.3.3 Electromyography

As facial EMG analysis provides an opportunity to explore both positive and negative

valence, this measure appeared uniquely applicable to investigating both optimal and sub-optimal

play experiences. Originally, three sites were chosen for EMG analysis: the OO and ZM, both

associated with enjoyment and positive valence, and the CS, associated with frustration and

negative valence. Preliminary informal testing of the EMG measurement indicated that facial hair

would prove intrusive in obtaining ZM activity, requiring either the removal of facial hair—

discounted due to ethical considerations—or the exclusion of participants with facial hair. Due to

the prominence of facial hair among the potential participant pool, and the remaining presence of

a secondary measure of positive valence (OO), ZM was ultimately removed from the

methodology.

92


As with EDA, the EMG system employed was supplied by BIOPAC Systems Inc., and

consisted of an EMG amplifier, 8mm Ag-AgCl shielded electrodes of the cup variety and 19 mm

double-sided adhesive collars for adhesion to the skin. The electrode cups were filled with a non-

irritating, hypo-allergenic gel conductant (‘SignaGel’). Impedance between electrodes was

checked using a UFI Model 1089 MkIII Checktrode.

The set-up period for EMG was variable and dependent on the success of the initial set-

up, as determined by the impedance. If the Checktrode reported unsatisfactory impedance levels

(> 10kΩ), the electrodes would be removed and the skin preparation process restarted. Initial

informal testing found that the set-up period would not exceed 15 minutes in length.

3.3.3.4 Electrocardiography

The relationships identified between time pressure, stress, anxiety, and HRV and R-R

analysis were identified as potentially revealing in the investigation of optimal and sub-optimal

play experiences, and particularly pertinent in application to an action horror video game. The

inclusion of ECG provided a secondary measure of arousal. ECG was also chosen due to its rate

of employment within both play experience literature and development in biofeedback.

As with all other psychophysiological measurements except EEG, the ECG system was

supplied by BIOPAC Systems Inc. The system consisted of an ECG amplifier, and used the 8 mm

Ag-AgCl shielded electrodes, collars and conductant gel used in EMG set-up. As with EMG,

impedance between the electrodes was tested using the MkIII Checktrode.

The set-up period, although again dependant on achieving an impedance level of less than

10kΩ, did not exceed 10 minutes during informal testing. If acceptable impedance was obtained

on the first application, the set-up would typically not exceed five minutes in length.

93


HR and HRV analyses, measures of arousal and both psychological and physiological

stress were chosen for the evaluation of ECG data. HR analysis in particular has represented a

locus of research in player experience research (Ballard & West, 1996; Drachen et al., 2010;

Fleming & Rickwood, 2001; Harmat et al., 2015; Kneer et al., 2015; Lim & Reeves, 2010;

Mandryk et al., 2004, 2006b; Nacke et al., 2011; Ravaja et al., 2006; Weber et al., 2009; Wu &

Lin, 2011), and as such is situated as a valuable analysis approach within the space. While HRV

is not typically assessed to the same extent, it may allow for a more nuanced understanding of

psychophysiological stress in player experience analysis (Keller & Bless., 2008; Harmat et al.,

2015; Russoniello et al., 2009). For this research, the peaks of the HF component—a direct

measure of cardiac parasympathetic activity—of HRV were assessed. While LF/HF and LF values

were obtained, they were not evaluated for this program of research due to the limitations of LF

analysis as a measure of sympathetic activity (Houle & Billman, 1999).

3.3.4 Selection of a Video Game Artefact

It was crucial that the video game artefact selected would support the aim, research

questions and conditions of the research program. Several requirements were established for the

selection of an appropriate video game for the study, as follows:

(i) representative of contemporary successful commercial video games and likely to

induce flow (and later, when flow was revealed as having possible issues, as

described in sections 3.4.4 and 4.3, likely to induce additional states such as

enjoyment, presence, autonomy and competence)

(ii) customisable—alterable to the extent of being able to develop both flow (optimal)

and low-flow (sub-optimal) game levels; this requirement also restricted the game

artefact to PC, due to the limited availability of ‘modding’ for console games

94


(iii) allowing for uninterrupted eight-minute play sessions to ensure adequate time for

a tonic psychophysiological experience to emerge

(iv) allowing players to ‘jump in’—no prior investment in the game required (players

did not have to be familiar with the game narrative, or to have completed a certain

amount of the game)

(v) sufficiently intuitive to be enjoyed by participants of all skill and experience

levels following their exposure to a short tutorial.

On the basis of these requirements, the video game chosen was Valve Corporation’s first-

person shooter zombie horror Left 4 Dead 2 (2009) (see Figure 15). Left 4 Dead 2 primarily

features gun-based combat with enemy zombies in a post-apocalyptic setting, accompanied by

three AI teammates (replaced with humans in multiplayer), with additional mechanics throughout

that require players to complete certain tasks—for example, fuelling a generator to lower a bridge.

As well as standard zombie enemies, the game also features ‘special enemies’—zombies with

unique abilities that can incapacitate players and require strategy to defeat.

Figure 15. Screenshot of Left 4 Dead 2 (Valve, 2009).

95


With more than four million copies sold in the year of its release (Chiang, 2011) and an

average rating of 89/100 on Metacritic, a review aggregator website often referenced in games

literature (Baldwin, Johnson, & Wyeth, 2014; Phillips, Johnson, & Wyeth, 2013), it was

determined that Left 4 Dead 2 was representative of a contemporary successful commercial video

game. Furthermore, Left 4 Dead 2 was heavily modifiable, featuring a developer-supported

‘modding’ tool and source kit and an established modding community. Initial testing of levels

within the game revealed several potential candidates for a minimum eight-minute play session.

Finally, the intuitive and recognisable nature of the game—a first-person shooter in which the

players must shoot zombie enemies—allowed players ‘jump in’ without extensive prior

experience, assisted by help tips that would explain objects or events as they were first

encountered.

Left 4 Dead 2 was also chosen due to its native inclusion of DDA in the form of an entity

known as the ‘AI Director’. DDA provides real-time adjustment of a game’s difficulty in response

to player status to ensure challenge‒skill balance, and has proven successful for invoking flow in

prior studies (Keller & Bless, 2008). It also allows for players of differing skill levels and game

familiarity to enjoy similar play experiences, as the optimal play condition adjusts automatically

to player ability. As the AI director operates primarily by manipulating the environment

and resources available to the player, this form of DDA aligns with the implementation

suggested by Negini et al. (2006).

3.3.5 Design Phase One

Initial artefact design followed the approach undertaken by Nacke and Lindley (2008).

The Boredom condition incorporated a linear level with weak opponents and repeating textures,

no winning condition, a limited choice of weapons, a high amount of health and ammunition

96


supplies, and a lack of surprises (enemies would not hide or attack from behind). To achieve

repeating textures and models within a linear level, the map was designed in the form of a simple

corridor without environmental clutter. Character conversations (friendly AI engaging in chatter

with one another that reveals story) were removed. Other changes from the standard game to

facilitate boredom focused on manipulating game challenge, and include the following:

(i) enemy AI altered to react slowly, or occasionally not at all, to player presence

(ii) no special zombies (powerful zombies with unique abilities present in the default

game)

(iii) no DDA

(iv) notably diminished enemy health, in that the player need only shoot them once

(v) player health unable to drop below 90%

(vi) high amount of ammunition and health pack supplies

(vii) only one available gun

(viii) no ‘winning condition’—the corridor does not end.

The Balance (flow) condition required very little modification to the original game, but

the game was nonetheless altered to remove what early pre-testing revealed to be luck-based game

changers. These were two ‘boss’ type zombies, known within the game as the ‘Witch’ and the

‘Tank’. The Witch (see Figure 16) had a chance of spawning at any time, or not at all, in any game

level; interaction with the Witch also required prerequisite knowledge that the novice players may

not intuit. Should the player not adopt the appropriate strategy (approaching quietly from behind,

or avoiding entirely), the Witch was capable of killing the player and all AI teammates, thus

prematurely ending the play condition.

97


The Tank (see Figure 16), while less lethal and more intuitive than the Witch, was a

disruptive event capable of spawning multiple times per play condition, dependent on player

performance. As a ‘mini-boss’, the time resources required to dispatch the Tank were

considerable. Due to the semi-random number of spawns, consistent play experiences could not

be guaranteed between participants and thus necessitated removal.

With the exception of the removal of these enemy types, the Balance condition was

unchanged from the version found in the official campaign. To this end, game difficulty was set

to ‘normal’ in the map editor. Common enemy zombies had 50 health, and would spawn in herd

sizes respective to the player performance, as judged by the AI Director. The player and their AI-

controlled teammates had 100 points of health each, and would take 2 damage per hit to their front

and 1 damage per hit to their back. The AI Director (DDA) was enabled. See Figure 18 for a

screenshot of the typical play experience for the Balance condition. The map level ‘The Parish’, a

winding maze-like cemetery, was chosen due to its high average completion time, under the

assumption that players would not be able to complete the level before the allotted play session

time expired.

Figure 16. Left 4 Dead 2 (Valve, 2009): Tank and Witch boss enemies.

98


3.3.6 Design Phase One: Artefact Design Flaws

Both conditions were rejected after initial informal pilot testing among colleagues and

collaborators revealed that flow was unnecessarily confounded with aesthetic quality, and that the

desired playtimes were not being achieved. While challenge‒skill imbalance was achieved

through direct manipulation of combat difficulty, aesthetic experience was compromised by the

removal of environmental assets in the Boredom condition. As it would be difficult to separate the

effect of reduced flow from that of reduced aesthetic quality, this condition was discarded in favour

of developing a boredom condition derived only from challenge‒skill imbalance (low flow), with

all environmental assets and character chatter intact. The Balance condition revealed errors in map

selection, as expert players were able to complete the map in less than the required playtime of

eight minutes and novice players easily became lost.

3.3.7 Design Phase Two

The second attempt at artefact design followed the framework established by Nacke and

Lindley (2008) only in regards to the manipulation of challenge‒skill balance through combat

difficulty; Keller and Bless’ (2008) singular focus on challenge‒skill balance largely provided the

framework for this design phase. Assuming challenge‒skill balance to be a central antecedent of

flow, all conditions directly manipulated this balance. To avoid confounds associated with

aesthetic differences, or potential differences in experiences if exposed to different environments,

all play conditions now took place within the same map. The map selected was the ‘The Port’, in

which players are required to fetch 16 canisters scattered throughout the map. Initial informal pilot

testing revealed that players were unable to complete the condition in fewer than eight minutes for

any condition, thus ensuring that no participant encountered a win condition.

99


With the reintroduction of aesthetic elements to boredom in the form of a new map (‘The

Port’), the Boredom condition now featured the same environmental assets, textures and character

conversations as the Balance condition. The Boredom condition continued to use all challenge

manipulations detailed in section 3.3.5, with the exception of the introduction of the gas canister

collection task. As the gas canisters are highlighted in the world in all conditions, rendering them

easy to locate, the task of collecting the canisters was not inherently challenging; additionally, the

large size of the map and the distribution of the canisters ensured a repetitive experience. For an

example of the typical Boredom condition experience, see Figure 17. The simple level design of

‘The Port’ made navigation considerably easier.

As with the first artefact design, the majority of level manipulation was undertaken to reduce

the likelihood of participants’ experiencing flow in the low-flow conditions; due to the presence

of DDA, it was expected that the game was capable of inducing flow without modification in the

Balance condition. The Tank was reintroduced to the Balance condition due to technical

limitations; while the Tank could be prevented from spawning, testing revealed that the game

Figure 17. Screenshot of Boredom condition (second iteration).

100


would still play the associated music and sounds. The Tank was thus introduced in a much

diminished capacity—whereas defaults Tanks possess 4000 health, the modified Tanks for this

condition had only 1000 (five times that of a common enemy type) and were easily dispatched by

the player. See Figure 18 for a screenshot of the typical play experience within the Balance

condition.

Figure 18. Screenshot of Balance condition.

The second design phase saw the addition of the Overload condition. The inclusion of this

condition enables a complete evaluation of the challenge‒skill balance spectrum: the measurement

of skill > challenge (Boredom), matched challenge‒skill (Balance), and challenge > skill

(Overload). To this end, game difficulty was set to ‘expert’ in the map editor. Common enemy

health was raised from 50 (in the Balance condition) to 1000. The player and their AI-controlled

teammates had 100 points of health each, and would take 20 damage per hit to their front and 10

damage per hit to their back. All special enemies, including the ‘Tank’, had their health multiplied

by four; this gave Tanks 16,000 health points. In addition to this, zombies were more likely to

spawn behind the player—there was no game-enforced limit on how many zombies could be

101


present in the map at a time, and the special infected spawn rate was increased. DDA was disabled

for this condition. See Figure 19 for a screenshot of the typical play experience for the Overload

condition. For an overview of the condition manipulations in Design Phase 2, see Table 3.

Figure 19. Screenshot of Overload condition.

Table 3. Game condition differences (second iteration)

Some features of Left 4 Dead 2 were altered for all conditions to preserve experimental

integrity and inhibit potential confounds. These alterations include the following:

The achievement system was disabled to ensure that no players experienced a reward for

achieving something that other players did not (for example, killing five ‘Hunter’-type

special enemies).

Boredom Balance Overload

Common

enemies

Impeded Standard Extreme

Special enemies Disabled Standard Extreme

Canister

collection

Enabled Enabled Enabled

DDA Disabled Enabled Disabled

102


Sprays, player-chosen images that can be ‘pasted’ into the game world by pressing a

button, were also disabled.

Players were restricted to only one of four playable characters (‘Nick’).

Weapon choice was eliminated from the game. In the default game, players may choose

to play with a sniper rifle, machete, assault rifle, chainsaw and so on. To ensure a similar

experience across all conditions and experiments, only the assault rifle as the primary

weapon and the pistol as the secondary weapon were enabled.

3.3.8 Tutorial

Prior to pilot testing of the second methodology design, a tutorial was created for the

purposes of introducing participants to the game mechanics, rules and environments. This was

undertaken with the intent of minimising the influence of learning effect and orienting response

(Stern et al., 2001), familiarising participants with the control scheme and input requirements of

the game, and reducing uncertainty in the tested play conditions.

The tutorial was developed within same map (‘The Port’) as the play conditions to maintain

consistency between game exposures. Pop-up speech bubbles throughout the tutorial were inserted

to guide participants throughout the map and provide graphic instructions for which keys were

needed for certain actions (see Figure 20). Throughout the tutorial, participants were introduced

(in order) to the following:

the actions required to pick up and equip a gun, ammunition and health pack

a single non-aggressive enemy, with instructions for how to shoot it

how to pick up fuel canisters, and pour their contents into the generator (the objective

within all play conditions)

how to track fuel canister silhouettes through the map

103


a single aggressive special enemy (‘Hunter’), with instructions for how to shoot it

time allotted for idle exploration, so as to allow a further period of acclimation to the

environment and controls.

The tutorial was scripted to ensure that the participant could complete all required actions

in fewer than five minutes, and allow a further two minutes for general exploration of the map. To

adequately guide the participant through the tutorial, some areas that were otherwise open to

participants in the play conditions were restricted through the insertion of fences; the only way to

access these regions was through a single path that required the completion of all tutorial

objectives.

Figure 20. Screenshot of tutorial.

104


3.4 STAGE 2—STUDY 1 (PILOT) AND REVISION

Pilot test of refined video game artefact; further modification to Boredom condition; exploration

of subjective constructs expanded beyond flow; development of additional experiment software

3.4.1 Introduction

The second phase of artefact design was piloted in an adaptation of the final proposed

methodology, without psychophysiological measurement. Psychophysiological measurement was

excluded from the pilot test, as the primary objective was to determine the success of the level

design in both achieving the optimal experience of flow (as in the Balance condition), and

inhibiting it (as in the Boredom and Overload conditions). After each condition, participants

answered self-report surveys addressing flow. Successful invocation of flow, as gauged by the

validated Long Flow State Scale (FSS-2), would allow for the exploration of RQ1a and RQ1a.

3.4.2 Methodology

The final methodology featured a within-subjects study design, employing semi-

counterbalanced video game artefacts to minimise order effect. Self-reported flow was evaluated

as a method for determining the conditions’ successes in both inhibiting and disinhibiting the

optimal play experience of flow, as well as the conditions’ rigour in manipulating challenge‒skill

balance and imbalance.

3.4.2.1 Semi-Counterbalanced Approach

While the study design employed some counterbalancing to control for order effects, the

conditions were not fully counterbalanced. The decision was ultimately predicated on the risk to

prolonged emotional or physiological response to the Overload condition; early internal play-

testing within Design Phase 2 (section 3.4) revealed the condition to be frustrating or

105


overwhelming, potentially influencing participant mood and receptiveness to subsequent

conditions. The limitations of this approach are discussed in section 6.5.

3.4.3 Study

The study took place between August and September 2014, and data was collected from

20 participants. The results from this study, discussed in Chapter 4, informed further design

revisions and refinements for the game artefact, experiment procedure and study design. These

iterations are discussed in the following section.

3.4.4 Revisions to Methodology

As described in section 4.3, the Overload condition performed as expected in terms of

showing lower levels of flow. However, expected differences between the Balance and Boredom

conditions were absent, prompting a reconsideration of the existing methodology. The program of

research was adapted to these findings in three ways:

further modification of the Boredom condition to remove existing potential for

engagement or enjoyment through challenge‒skill manipulation

refocusing of the psychological constructs investigated in the program of research; a

widening of scope allows for the investigation of other psychological constructs, as

identified within games research literature, beyond flow

while flow is still to be evaluated, its role is to be pared down.

3.4.4.1 Continued Modification of the Boredom Condition

The final iteration of the Boredom condition removed enemies and combat entirely; the

gameplay consisted exclusively of retrieving the gas canisters scattered throughout the map (see

Figure 21). Despite the removal of combat altogether, the inclusion of canister collection and

travel in the virtual world ensured that the game remained sufficiently game-like.

106


It was possible that the (limited) inclusion of zombie targets, despite their lack of threat

to the player, was still sufficiently engaging as to inhibit the potential for boredom within the

condition. As discussed in sections 4.3.1 and 4.3.2., it was also possible that the anticipation of

future enemies may have prevented disengagement from the game. In an effort to ensure the

Boredom condition was capable of inducing boredom, the combat in the game was removed

altogether.

The various iterations of the video game artefact conditions are detailed in Table 4. Please

note that the Overload condition was not introduced until Iteration Two. For video of these

conditions, refer to Appendix G.

Figure 21. Boredom condition (second iteration), feat. no combat.

107


Table 4. Game condition differences (third iteration)

Iteration One Boredom Balance Overload

Map Custom ‘corridor’ The Parish -

Common enemies Impeded Standard -

Special enemies Disabled Standard -

Tank Disabled Disabled -

Canister collection - - -

DDA Disabled Enabled -

Iteration Two Boredom Balance Overload

Map The Port The Port The Port

Common enemies Impeded Standard Extreme


Tank Disabled Impeded Extreme

Canister collection Enabled Enabled Enabled


Iteration Three Boredom Balance Overload

Map The Port The Port The Port

Common enemies Disabled Standard Extreme


Tank Disabled Impeded Extreme

Canister collection Enabled Enabled Enabled


108


3.4.4.2 Refocusing of Psychological Constructs

As discussed in section 4.3, it is also possible that low flow is either a) inherently difficult

to obtain in immersive, aesthetic video game conditions, b) not reliant on the obtainment of

challenge‒skill balance or c) not able to be accurately measured by the FSS-2 within this context,

particularly where the skills of the player exceed the challenges of the game. It was thus decided

to withdraw from flow as the sole psychological construct for psychophysiological evaluation;

instead, the optimal and sub-optimal play experiences—as obtained through optimal and sub-

optimal challenge‒skill balance—would also provide a lens through which to examine other

psychological constructs of note in games literature. The scope of the research program was thus

broadened to include evaluation of presence/immersion, enjoyment, affect and features of SDT,

as well as flow (in a reduced capacity).

3.4.5 Development of Sequencing Software

The development of sequencing software was undertaken prior to the commencement of

data collection for the final study. This software, dubbed ‘Sequencer’, was developed and designed

in collaboration with a colleague in QUT’s Computer‒Human Interaction department. The

development of Sequencer was undertaken with the primary aim of reducing participant‒

researcher interaction, identified by Mandryk et al. (2006) as a potential confound in the recording

of psychophysiological measures.

Sequencer is an application that makes use of Microsoft’s .NET functionality. The

program’s primary functionality is that it can follow event branches based on conditioned logic

(i.e., the software takes a coded participant number and follows a sequence of steps associated

with that condition). Before each experiment, the experiment coordinator can input a participant

ID and set up the program for the participant front-end. Using the participant number, the

109


experiment sequencer determines which branch of the experiment to run and loads surveys, game

environments and baseline applications in the coded sequence. From the launched state, the

participants can commence the study by clicking on simple ‘Next’ buttons. Unlike traditional

Windows applications, the environment obfuscates the task bar of the Windows environment to

intentionally remove participant access to the system clock, and to prevent participants from

accidentally exiting the software environment. A secondary feature of the program is inbuilt timers

that control the length of the game conditions and the baseline application, ensuring an identical

playtime across participants.

While Sequencer was successful in minimising participant‒researcher interaction, it also

had several secondary benefits. In the absence of the program, the researcher would have had to

necessarily launch all programs and surveys themselves, thus causing delaying marking the

initiation of experiment events in the psychophysiological recording software (or an additional

member of the research team would have needed to be present to assist in this task). It also

maximised physical distance, allowing the researcher to sit at a machine located beyond

participant eyesight; the participant may thus have been less likely to feel as though their video

game play and performance were being watched, which potentially influenced their physiological

response. Sequencer had the additional benefit of reducing the chance for human error during

experiment runtime. The complex nature of the procedure (involving three computers, five

programs and five psychophysiological measures), and the requirement for precision, were at

odds, and so benefitted from the inclusion of the Sequencer software. Finally, Sequencer had the

ability to gracefully recover from an unexpected error in experiment runtime (for example, a

computer crash); in the event of abrupt experiment failure, correct order was resumed through the

manual selection of the last completed sequence of the experiment. See Figure 22 for a screenshot

of the Sequencer software’s main menu.

110


3.4.5.1 Baseline Application

A simple application that renders the entirety of the computer screen black was developed

as the ‘task’ for the psychophysiological baseline. Baselines were timed to terminate at one minute

and 30 seconds. This application was launched by Sequencer throughout the experimental

procedure.

Figure 22. Screenshot of Sequencer main menu.

111


3.5 STAGE 3—STUDY 2

Final study design and methodology; data collection

3.5.1 Introduction

The final iteration of the video game conditions and experimental methodology was

deployed for data collection over a period of 13 months. This final methodology allowed for the

exploration of both the aim of the research program, and the investigation of research questions

RQ1a and RQ1b. The expansion to include multiple subjective constructs beyond flow enabled a

more nuanced exploration of the relationship between psychophysiological and subjective

response; furthermore, the iteration of both the video game artefacts (conditions and tutorial) and

assistive experimental software (Sequencer and baseline program) allowed for a robust study

design that supported the psychophysiological method by minimising participant‒researcher

contact and automation of timed events.

3.5.2 Methodology

The final methodology featured a repeated-measures within-subjects study design,

employing semi-counterbalanced video game artefacts to minimise order effect. A

psychophysiological and subjective approach was undertaken through use of both the subjective‒

quantitative survey method and objective‒quantitative psychophysiological method.

Throughout the study, participants’ subjective responses were obtained through the use of

three validated scales: the PENS, the Short Flow State Scale (S FSS-2), the interest/enjoyment

subscale of the Intrinsic Motivation Inventory (IMI), the Self-Assessment Manikin (SAM) and the

Positive and Negative Affect Schedule (PANAS). Two custom items were also included to ask

plain questions such as, ‘How fun did you find this session?’ and ‘How did your ability to play

the game match the challenges of the game?’ For this thesis, analysis of subjective measures was

112


limited to the PENS, S FSS-2 and IMI scales. Finally, five psychophysiological measures were

used to obtain cortical, cardiovascular, sweat gland and facial muscle activity. For a complete

discussion of the method and measures used, see Chapter 5.

3.6 ETHICS AND LIMITATIONS

Both studies in this program of research were evaluated as low risk, with no added risk

beyond standard interaction with video games and technology. Participants were informed of the

game’s classification category (MA15+) and content prior to participating in the study in the

recruitment materials, participant information consent sheet and pre-experiment briefing. This

allowed for those uncomfortable with first-person shooters to not participate or otherwise rescind

interest. Furthermore, the participant pool was limited to those aged 17+ in consideration of the

game’s classification, and to best satisfy low-risk human ethics requirements at QUT. Due to

concerns for the consistency of data gathered from ECG, individuals with a history of heart

arrhythmia were selected against and thus requested to not participate in the recruitment materials.

While there was a chance that participants would feel tension or discomfort during the set-

up of the psychophysiological measures, they were informed as to the nature of the physiological

measurements in the same recruitment and information materials. Any potential participants who

may have felt uncomfortable with the concept of psychophysiological measurement were thus

granted the opportunity at multiple stages to not participate or remove themselves from the study.

Participants were also advised to alert the researcher to any discomfort they experienced

throughout the study, and were informed during the pre-experiment briefing that they could

withdraw at any time.

113


Finally, the use of hypoallergenic soaps and gels throughout the study minimised the chance

of allergic reactions. All electrodes were thoroughly cleaned, disinfected and rinsed between

experiment sessions for the dual purposes of hygiene and equipment maintenance.All responses

and data were anonymised and stored in secure environments, accessible only on password-

protected computers. This data was made available only to the researchers involved in the study,

as detailed in participant information consent sheets. The signed information consent sheets were

stored in a lockable cabinet on university property.

Both studies were approved by the QUT University Human Research Ethics Committee

(approval #1300000796). The second full study was granted approval as an amendment to the

original documentation submitted for the pilot study.

114


3.7 STAGE 4—ANALYSIS AND INTERPRETATION

Data treatment, analysis and interpretation.

3.7.1 Introduction

The final step of this research program was the treatment, analysis and interpretation of all

psychophysiological and subjective data. The results were intended to underpin the

methodological and knowledge contributions made to player experience research, and allow for

the comprehensive exploration of RQ1a and RQ1b.

3.7.2 Scope

Upon completion of data collection, all psychophysiological measures were treated, cleaned

and analysed in compliance with approaches detailed in sections 2.8.7 and 5.1.8. This represented

a notable time investment within this research program; associated implications for using

psychophysiological measures in assessing player experience are discussed in section 6.4.1.

Analysis was then undertaken in a series of one-way MANOVAs and ANOVAs, revealing

significant differences between the optimal and sub-optimal conditions across both subjective and

psychophysiological measures. The results confirm the success of the conditions in the creation

of optimal and sub-optimal conditions, informing the interpretation of the psychophysiological

results. The results are reported in sections 5.2 and 5.3.

The interpretation of results was undertaken through the lens of both player experience and

psychophysiological research; while player experience components offered reasonable

explanations for many results revealed for the physiological responses, some results may have

emerged as a consequence of basic psychophysiological principles (e.g., habituation). The strength

of the results is also investigated through effect sizes—see sections 5.5 and 5.6 for these

discussions. The results are assessed in terms of the thesis aim and research questions in section

115


6.3. The employment of psychophysiological evaluation within this program of research is

critically assessed, allowing for the identification of research limitations, in section 6.5. In sections

6.4 and 6.5, recommendations are made for psychophysiological methodologies in player

experience research with an eye to future research opportunities.

116


4 STUDY 1: PILOT

Study 1 was undertaken to explore the effectiveness of the play condition artefacts in

inhibiting and disinhibiting the experience of flow. A three-condition approach allowed for the

investigation of flow through challenge‒skill balance, low flow through challenge > skill

imbalance (frustration/anxiety) and low flow through skill > challenge imbalance (boredom). The

successful development of three conditions matching these criteria enabled a comprehensive

psychophysiological exploration of a concept synonymous with optimal experiences within both

player experience and psychological research spaces (Sweetser & Wyeth, 2005; Csikszentmihalyi,

1990).

It was expected that the results from this preliminary study would inform ongoing design of

the play conditions and experiment procedure, as well as the broader paradigms of the research

program. The culmination of these developments facilitated the commencement of the final study,

including psychophysiological measures, and enabled the consideration and exploration of RQs

1, 1a and 1b.

4.1 METHOD

4.1.1 Recruitment

The desired sample size for the pilot study was 20 participants. This offered opportunity

for the preliminary analysis and evaluation of the subjective response to condition, as well as

insight into the general success of the three play conditions in either inhibiting or disinhibiting the

optimal play experience through flow and challenge‒skill balance manipulation. Recruited

individuals were aged 17 and older (see section 3.6). Participants were recruited from the Bachelor

117


of Games and Interactive Entertainment cohort at QUT and through snowball sampling.

Compensation for participation was a game software key, which granted free access to a library

of games developed by Valve Corporation, and the opportunity to enter a draw to win a $50 gift

voucher.

4.1.2 Measures

4.1.2.1 Demographics Questionnaire

The demographics questionnaire (see Appendix D) included age, gender, level of

experience with video games, level of experience with first-person shooters and estimated number

of hours spent playing Left 4 Dead 2 and 1 (with no prior experience with the game required for

participation in the study).

4.1.2.2 Long Flow State Scale

The FSS-2 (Jackson & Eklund, 2002) is a 36-item survey measured on a 5-point Likert

scale, with ‘1’ representing ‘strongly disagree’ and ‘5’ representing ‘strongly agree’. The scale

consists of subscales (four items each) measuring each of the eight components of flow—skill,

concentration, clear goals, unambiguous feedback, action‒awareness, sense of control, loss of self-

consciousness and transformation of time—and a subscale measuring the autotelic experience

associated with flow. Scores are calculated for each of the nine subscales and for total flow (the

nine subscales combined). The FSS-2 is a validated measure for evaluating the experience of flow

in various settings, and has previously been successfully applied to video (Kivikangas, 2006;

Vella, Johnson, & Hides, 2013; Harmat, 2015). See Appendix A for sample items.

118


4.1.3 Laboratory

The experiments took place within a temperature-controlled computer laboratory at QUT.

Each computer within the laboratory is divided by partitions into an individual ‘booth’, providing

privacy in the use of each machine. The computer used in the experiment is a corner booth, chosen

so that the participant is unable to be seen (or unable to see) the outside hallway through the door

window. The researcher sits at a machine behind and diagonal to the participant, and does not face

the participant’s screen, again encouraging privacy in the participant’s use of the machine. The

laboratory is lit artificially, with no outside-facing windows that would introduce natural lighting.

During data collection, only the participant and the researcher were present in the laboratory. See

Figure 23 for the laboratory configuration.

4.1.4 Procedure

The experimental sessions took place in the computer laboratory detailed in section 4.1.3,

with only one participant tested per one-hour session. Each experiment session took place between

10 am and 6 pm, with most sessions occurring on weekdays. After providing informed consent,

Figure 23. Experimental laboratory. Left: participant desk, feat. partitions; right: view of participant desk from

researcher desk.

119


participants were advised to be aware of a copy of the Left 4 Dead 2 control scheme mounted on

the partition walls.

Participants were then directed to answer questions regarding demographics and previous

experience with both video games in general and Left 4 Dead 2 specifically. Once this was

completed, participants then played a four-minute structured tutorial for the game that exposed

them to the mechanics and requirements they needed to be aware of for gameplay. Finally,

participants played three 10-minute game sessions in semi-counterbalanced order

(Boredom/Balance/Overload, or Balance/Boredom/Overload), with a five-minute questionnaire

segment punctuating each game condition. Once the experiment was completed, participants were

debriefed and thanked for their time. See Figure 24 for a depiction of the experiment procedure.

Introduction00.00 - 03.00

Background Survey

03.00 - 6.00

Control Scheme Diagram

6.00 - 7.00

Tutorial7.00 - 11.00

First Play Session

(Bore/Bal)11.00 - 21.00

First FSS-221.00 - 26.00

Second Play Session

(Bore/Bal)26.00 - 36.00

Second FSS-236.00 - 41.00

Third Play Session

(Overload)41.00 - 51.00

Third FSS-251.00 - 56.00

Debriefed and Thanked

56.00 - 60.00

Figure 24. Study 1: Experiment procedure.

120


4.1.5 Participants

Twenty participants participated in the study, 95% male and aged between 17 and 31

(mean = 20.2, SD = 3.24). On a Likert scale of 1–7, with ‘7’ representing ‘extremely experienced’

and ‘1’ ‘not at all experienced’, participants self-rated as a mean of 6 (SD = 1.12) for ‘general

experience with video games’ and 5.3 (SD = 1.56) for ‘experience with first-person shooters’.

Participants reported between one and 80 hours spent playing video games per week, with a mean

of 25.45 hours (median = 20, SD = 20.05). In terms of familiarity with Left 4 Dead franchise,

participants were asked how many hours they had spent playing both Left 4 Dead 1 and Left 4

Dead 2. Eleven participants had never played Left 4 Dead 1 before; the remaining nine had played

between 1 and 100 hours (mean = 20.78, median = 10, SD = 30.93). In terms of Left 4 Dead 2, 11

had never played before, with the remaining 9 having played between 2 and 150 hours (mean =

32.56, median = 10, SD = 48.84). Overall, 60% of participants had previously played a Left 4

Dead title.

4.2 FINDINGS

A within-subjects MANOVA was conducted using gameplay condition (Boredom,

Balance or Overload) as the independent variable and all outcome measures (the nine subscales

as well as total flow) as dependent variables. All statistical assumptions of MANOVA were met,

with the exception of univariate outliers identified on the challenge‒skill balance and

transformation of time subscales from a single participant. No substantive differences in results

were found with outliers removed, and in the interest of statistical power, the results reported here

include all cases. Using Wilk’s Lambda revealed a statistically significant effect of condition on

the combined dependent variables (Λ = 0.259, F(18,60) = 3.221, p < 0.005; partial η2 = .491).

Univariate follow-ups revealed differences in terms of challenge‒skill balance (F(2,38) = 13.744,

121


p < .005), merging of action and awareness (F(2,38) = 13.744, p = .007), and sense of control

(F(2,38) = 4.552, p = .017). Total flow was also found to differ significantly between conditions

(F(2,38) = 4.867, p = .013).

Pairwise follow-up tests using Bonferroni corrections were conducted on these three

subscales and total flow. Challenge‒skill balance was significantly higher for the Boredom

condition than for the Overload condition (p = .008); it was also significantly higher for the

Balance condition than for the Overload condition (p < .005). The Boredom condition also scored

significantly higher than the Overload condition in the merging of action and awareness (p = .005)

and sense of control subscales (p = .046). Finally, for total flow, no difference between the

Boredom condition and the Balance condition was identified (p = .097), but the Balance condition

revealed significantly greater flow than the Overload condition (p = .017). For a visualisation of

these results, see Figures 25 and 26. Overall, the results show that greater total flow was

experienced in the Boredom condition than in the Overload condition. This difference seems to be

a function of participants reporting relatively high levels of challenge‒skill balance, merging of

action‒awareness and sense of control in the Boredom condition. Additionally, no significant

differences were found between the Boredom condition and the Balance condition.

122


Figure 27. Study 1: Flow State Scale subscale results

Figure 28. Study 1: Flow State Scale total flow results.

123


4.3 DISCUSSION

4.3.1 Difficulties in Reducing Flow in Immersive Games

A proposed explanation for the absence of differences in flow experiences between the

Boredom and Balance conditions is the potential for flow experience regardless of challenge‒skill

imbalance in immersive video games, suggesting that a challenge‒skill imbalance does not

necessarily negate flow in immersive video games. As Left 4 Dead 2 features a highly detailed

environment, it is possible that some participants may have derived aspects of flow (such as altered

perception of time, focused concentration and loss of self- consciousness) from world exploration.

In other words, regardless of very low levels of challenge offered by the enemies in the game,

participants were able to achieve flow by exploring and/or enjoying the aesthetic qualities of the

game world. As many commercial titles feature detailed environments, this may point to issues

with the analysis of flow experienced in video games.

Expanding on the concept of engagement through immersion or aesthetic quality, it is also

possible that the challenge > skill imbalance achieved by the Overload condition was successful

in preventing flow by disinhibiting immersion within the world. This may be due to repeated

player deaths, limited mobility (constrained by enemies) and reduced chance for exploration.

The potential challenges of reducing flow in immersive games highlighted risks in the

exclusive measurement of flow as a tool for psychophysiological comparison. Therefore, the

decision was made to expand the program of research to assess other psychological constructs

associated with the player experience; for further discussion of this, see section 3.4.4.

124


4.3.2 Unsuccessful Condition Design

The absence of any discernible differences in the flow experience between the Boredom

and Balance conditions may also indicate errors in condition design. It may be that participants

were still sufficiently challenged by the demands of the Boredom condition, despite the range of

constraints introduced to ensure very low levels of challenge. Specifically, player health never fell

below 90%, enemies would die from a single shot, no ‘special infected’ would spawn and a low

enemy spawn rate was used (less than one-third of that seen in the flow condition). Regardless, it

may be that participants found unexpected ways to seek challenge in the Boredom condition—for

example, they may have felt challenged as a function of not being sure when enemies would appear

(the anticipation of enemies in some ways balancing the relative lack of enemies), or perhaps by

obtaining as many fuel canisters as possible before play time concluded in the condition

(confirmed anecdotally by two participants to the researcher). It is also possible that the presence

of combat was still perceived as sufficiently threatening or challenging by the players, regardless

of the reality of the challenge; this may have been influenced by anticipation of continued combat

or anticipation of elevating combat difficulty.

These conclusions directly prompted additional iterations to the play condition designs;

in particular, Boredom was revised to minimise the risk of perceived challenge. While this still

allowed the potential for player-created challenge, it was determined that such a possibility was

unavoidable within the context of computer games. The Boredom condition was predominately

redesigned to remove combat altogether; for extrapolation on this, see section 3.4.4.

4.3.3 Challenge‒Skill as an Antecedent

Keller and Landhäußer (2012) and Csikszentmihalyi et al. (2005) identify challenge‒skill

balance as an antecedent of flow. However, Fong et al. (2014) note that this balance between

125


challenge and skill may be commonly associated with, but not always necessary for, flow to occur.

It may thus be that the experience of flow is possible despite challenge‒skill imbalance; in this

case, sole focus on challenge‒skill balance may not have been adequate for successful flow

manipulation. Rheinberg et al. (2003) propose that flow is divisible into two factors: ‘absorption

by activity’ and ‘fluency of performance’. In this approach, flow through absorption is often

associated with balanced or slightly challenging activities, whereas flow through fluency is

stronger under low-challenge activities. It is possible that participants experienced flow through

absorption in the Balance condition, and flow through fluency in the Boredom condition.

As the methodology was revised to include the assessment of additional subjective player

experience constructs, the role of challenge‒skill balance as an antecedent to flow was no longer

essential for the program of research. The manipulation of challenge‒skill balance still provided

a useful tool for comparison among play experiences; furthermore, due to the importance of

challenge and challenge‒skill balance in the player experience, challenge‒skill manipulation

remained capable of promoting and inhibiting optimal experiences. For discussion on this, see

section 2.3.1.

4.3.4 Scale Applicability

By contrasting a flow experience with a boring experience, this study raises the question

of the FSS-2 scale’s applicability to some video game experiences. The FSS-2 is a commercial

scale, and specific scale items cannot be published; the relevant subscales as a whole are discussed

here.

Two of the subscales in question—the ‘merging of action and awareness’ subscale, and

the ‘sense of control’ subscale—present the possibility of answering the question of scale

applicability in a manner congruent with experiences of both flow and boredom or disengagement.

126


The ‘merging of action and awareness’ subscale contains items focused on performing actions

automatically and without much thought; the ‘sense of control’ subscale contains items focused

on feeling control over what one is doing. While these experiences are true of the flow experience,

they are also arguably true of a boring or unchallenging experience. As the Boredom condition

was not mentally taxing, it follows that participants did not need to put ‘too much thought’ into

their actions; similarly, as the game is not mechanically challenging and was selected for its

intuitiveness, it is likely that participants generally felt particularly confident in their control of

the Boredom condition. This aligns with Keller and Bless’s (2008) findings, in which the highest

levels of perceived control were reported for the boredom condition used in their study. It may be

that video games otherwise not likely to induce flow still offer participants the opportunity for

high levels of sense of control and merging of action‒awareness. In this way, the FSS-2 may

indicate high levels of flow in video games with these features.

It is particularly notable that no significant difference was found between the Balance and

Boredom conditions for the challenge‒skill balance subscale. A possible explanation stems from

the subscales including items that could be interpreted as asking if the respondent has sufficient

skills to meet the presented challenge. Participants could have sensed that their skills were enough

(or more than enough) to meet the challenges of the Boredom condition, leading to high scores on

this subscale. The results from the current study raise the possibility that the FSS-2, when applied

to boring or exceptionally easy video game scenarios, could result in high ratings of flow when

flow may not actually be occurring.

As in section 3.4.1, the potential limitations of using this measure were minimised by

shifting focus from the exclusive physiological assessment of flow. As limitations in the

applicability of the FSS-2 to sub-optimal game experiences were identified, the subjective

127


assessment of flow was pared down to the Jackson S FSS-2—an abridged nine-item version of the

FSS-2. This allowed for the ongoing psychophysiological assessment of flow, and the introduction

of additional subjective measures, without notably lengthening survey time commitments in the

experimental process.

4.4 CONCLUSIONS

The development and evaluation of the flow and low-flow experimental play conditions has

uncovered difficulties concerning operationalising and measuring flow in games research, and

raises questions regarding the role of challenge‒skill balance in flow. Study 1 highlights the

difficulties inherent in manipulating flow: as aesthetically ‘immersive’ experiences are common

attributes of commercial games, continued research must remain cognisant of this as a potential

confound with flow. As for the Boredom condition used in the current study, it was uncertain

whether people experienced flow regardless of the low challenge experienced in the level, or

whether higher challenge was created or perceived by participants in unexpected ways. Finally,

the applicability of Jackson et al.’s FSS-2 scales to play experiences—especially sub-optimal

experiences—is potentially limited, particularly in the context of challenge‒skill manipulation.

These findings informed the final iteration of the video game conditions and study methodology,

primarily in the redesign of the Boredom condition and the introduction of additional

psychological constructs.

128


5 STUDY 2—MAIN STUDY

Both chapters 3 and 4 have demonstrated the iterations in the design and development of

the final study. These steps were informed by recommended practice in the psychophysiological

space, insights gathered from contemporary player experience literature and findings gathered

from the preliminary pilot study. These iterations culminated in creating and refining study

software artefacts (play conditions, tutorial, Sequencer and baseline program), developing a robust

and relevant psychophysiological method and establishing a final subjective approach to allow for

a simultaneous biometric and psychometric investigation of the player experience. The

culmination of these developments allowed the final study to commence and to encompass

psychophysiological measures, thus enabling the consideration and exploration of the RQs 1, 1a

and 1b.

1. How effectively can the psychophysiological method be used to evaluate the player

experience?




measures, most reliably predict specific components of the player experience as assessed

by subjective measures?

Evaluating the psychophysiological response to the Boredom, Balance and Overload

conditions facilitated an exploration of differences in response, and a determination of which

physiological processes may be associated with optimal or sub-optimal play experiences, thus

addressing RQ1a. The concurrent analysis of the subjective experience—employing flow,

129


enjoyment, autonomy, competence and presence as the self-reported psychometrics for

evaluation—also enabled a manipulation check to determine the game conditions’ success in

evoking (or inhibiting) optimal play experiences.

Furthermore, exploring the psychophysiological response as a predictor of subjective

response in statistical analysis allowed for the investigation of RQ1b. This enabled a greater

understanding of components of the player experiences and which psychophysiological measures

most effectively predict specific components of the player experience. This, in turn, provided

guidance regarding these components in future research and commercial contexts.

5.1 METHOD

5.1.1 Recruitment

The desired sample size was 90 to 150 participants aged 17 and older, which facilitated

the analysis and evaluation of the psychophysiological and subjective response to condition, as

well as the analysis of psychophysiology as a predictor of subjective response via multiple

regression testing (Tabachnick & Fidell, 2007). This sample size in particular was chosen as it

would allow for inferential statistics such as multiple regression testing, which would allow insight

regarding the relative predictive power of different physiological measures. Participants were

recruited from the Bachelor of Games and Interactive Entertainment cohort at QUT, the wider

university undergraduate student base through targeted social media pages, the general public via

gaming forums and social media, and through snowball sampling. Compensation for participation

was a game software key, which granted free access to a library of games developed by Valve

Corporation, and the opportunity to enter a draw to win a $50 gift voucher.

130


5.1.2 Self-Report Measures

5.1.3 Demographics Questionnaire

The demographics questionnaire from the study undertaken in Stage 3 (section 4.1.2)

was reused for this study.

5.1.4 Short Flow State Scale

In lieu of Jackson and Eklund’s FSS-2 (2002) employed in Stage 2, the subjective focus

on flow was pared back to instead employ the S FSS-2 (Jackson & Eklund, 2002). This exchange

was undertaken in consideration of time constraints for the experimental procedure, and reflected

the shift in research program objectives from flow to multiple subjective states.

The S FSS-2 features a 9-item scale instead of the 36 items present in the previous scale.

Like the FSS-2, the S FSS-2 items are measured on a 5-point Likert scale with ‘1’ representing

‘strongly disagree’ and ‘5’ representing ‘strongly agree’. Each item evaluates one of the eight

components of flow—skill, concentration, clear goals, unambiguous feedback, action‒awareness,

sense of control, loss of self-consciousness and transformation of time—with an additional item

assigned to measuring the autotelic experience associated with flow.

5.1.5 Player Experience of Needs Satisfaction

The development of the PENS scale has allowed researchers to explore the three needs of

SDT (autonomy, competence and relatedness) through a measure uniquely developed for the

experience of video game play (Ryan et al., 2006). The PENS also evaluates presence and intuitive

controls, core influences on the player experience (Ryan et al., 2006), as separate subscales.

Presence, as discussed in section 2.5, is theorised to be positively associated with increased

intrinsic motivation. Intuitive controls assists in enabling feelings of competence, autonomy (by

131


not stymieing the player with awkward or difficult controls) and presence (by facilitating feelings

of ‘being there’ through to providing an intuitive control scheme that does not require active

consideration) (Ryan et al., 2006). The PENS survey thus features five distinct subscales:

autonomy, competence, relatedness, presence and intuitive controls.

The efficacy of the PENS survey was investigated across four studies, which found that

the measure was successful in determining needs satisfaction and play motivation (Ryan, Rigby,

& Przybylski, 2006). The PENS survey has since been employed and validated in digital games

literature in several contexts, including as support for investigating player wellbeing (Vella,

Johnson, & Hides, 2013) and exploring personality and game genre preferences (Johnson &

Gardner, 2010), and as a consistent tool for player experience analysis (Brühlmann & Schmid,

2015).

Three of the needs satisfaction subscales were included in the study design: competence,

autonomy and presence, which are thought to be core components of the player experience (Ryan,

Rigby, & Przybylski, 2006). The PENS scale has been shown to be a statistically reliable measure

(Johnson & Gardner, 2010); while it contains a subscale for relatedness, this was excluded from

the study on the grounds that the game is a single-player experience and no NPCs were present in

the game. Similarly, the intuitive controls subscale was excluded from the study, as the control

set-up was not manipulated in any way.

Items were rated on a Likert scale of 1–7 (‘7’ representing ‘strongly agree’), and are as

follows (also included in Appendix B):

Competence: ‘I felt very capable and effective when playing.’

Autonomy: ‘I did things in the game because they interested me.’

Presence: ‘I experienced feelings as deeply in the game as I have in real life.’

http://dl.acm.org/author_page.cfm?id=81557646656&coll=DL&dl=ACM&trk=0&cfid=871961143&cftoken=72277050

132


5.1.6 Intrinsic Motivation Inventory

Predicated on SDT, the IMI is a scale for measuring subjective experiences in relation to

an activity. It is considered a reliable measure of its subscales: interest/enjoyment, competence,

effort, value, pressure and perceived choice (autonomy). An additional subscale has since been

added, though has not yet been validated (relatedness). Of these subscales, only interest/enjoyment

is a direct measure of intrinsic motivation, and so features the greatest number of items (Ryan &

Deci, 2000). As the IMI and PENS are similarly rooted in SDT, it was determined that including

the needs satisfaction subscales from the IMI would be largely repetitious, and that those subscales

(competence, autonomy, relatedness) should not be included in the study design. However, the

interest/enjoyment subscale of the IMI does not have an analogue in the PENS, and was used to

gauge the interest and enjoyment of participants across the experimental conditions. Items were

rated on a Likert scale of 1–7 (‘7’ representing ‘very true’), and are as follows (see Appendix C

for all items):

‘I would describe this activity as very interesting.’

‘I enjoyed doing this activity very much.’

5.1.7 Psychophysiological Measures

The psychophysiological measures employed for this study—EMG (OO), EMG (CS),

EEG, EDA and ECG—were identified in section 3.3.3 and selected for their adherence to criteria

of accessibility, ease of deployment, high employment rate within games literature, applicability

to player experience evaluation, affordable temporal and financial costs, and complementarity

with one another. These measures were chosen early in the study design to ensure that the final

methodology would remain feasible and congruent with the psychophysiological measures’

133


limitations and requirements (e.g., as the instruments were largely non-mobile, participants would

not be able to move from their desks after the set-up period).

5.1.8 Ethics

Ethical approval was sought and granted by the QUT ethics committee (approval

#1300000796); see section 3.6

5.1.9 Procedure

The experimental sessions took place in the computer laboratory detailed in section 4.1.3,

with only one participant tested per two-hour session. Each experiment session took place between

10 am and 6 pm, with most sessions occurring on weekdays. After providing informed consent,

participants were given the opportunity to use the bathroom before the experiment session

commenced. Participants were then taken to a sink within the laboratory and asked to wash their

hands in preparation for the EDA electrodes. Upon drying their hands, participants were seated at

a PC in a corner booth.

The electrodes and instruments for the psychophysiological measures were then applied

over a duration of approximately 30 minutes. In some instances, application could take longer if

it was determined that the impedance between the electrodes was too high, or the electrodes were

not providing a clean physiological trace. EMG and ECG sites were appropriately prepared prior

to application with the use of an abrasion gel, gauze wipe and alcohol wipe. EEG electrodes were

saturated with saline prior to application.

Participants were then directed to answer questions regarding their demographics and

previous experience with both video games in general and Left 4 Dead 2 specifically. They then

played a four-minute structured tutorial for the game that exposed them to the mechanics and

requirements they would need to be aware of for gameplay. They then played three 10-minute-

134


and-30-second game sessions in semi-counterbalanced order (Boredom/Balance/Overload, or

Balance/Boredom/Overload). The Sequencer program would automatically deliver participants

the appropriate surveys after each session. A two-minute baseline was also delivered by Sequencer

at the start of the experiment, in between in each play session and after all the play sessions were

completed.

Once the experiment was completed, the psychophysiological equipment was removed

from the participant. Participants were offered a wet wipe to clean any remaining electrode gel

from their person, and were then verbally debriefed and thanked. See Figure 27 for a visual

summary of the experimental process.

135


5.1.10 Attachment of Psychophysiological Measures

All physiological measures were prepared and attached in accordance with standard

procedures in the psychophysiological space, as detailed in section 2.8.7. This section describes

the individual attachment processes employed in this study for each physiological measure.

Introduction00.00 - 05.00

Background Survey

05.00 - 10.00

Hands Washed

10.00 - 12.00

EDA Setup12.00 - 17.00

EMG (OO) Setup

17.00 - 24.00

EMG (CS) Setup

24.00 - 31.00

ECG Setup31.00 - 38.00

EEG Setup38.00 - 43.00

Re-application 43.00 - 50.00

Baseline50.00 - 52.00

Tutorial52.00 - 57.00

Baseline57.00 - 59.00

First Play Session

59.00 - 70.00

First Surveys70.00 - 77.00

Baseline77.00 - 79.00

Second Play Session

79.00 - 90.00

Second Surveys

90.00 - 97.00

Baseline97.00 - 99.00

Third Play Session

99.00 - 110.00

Third Surveys

110.00 - 117.00

Baseline117.00 - 119.00

Equipment Removed

119.00 - 124.00

Debriefed124.00 - 126.00

Figure 29. Experimental procedure.

136


5.1.11 Electrodermal Activity Attachment

As discussed in section 5.1.6, all participants washed their hands at a sink provided in the

laboratory prior to the attachment of the disposable EDA electrodes. Sink water was heated so to

prevent reactionary restriction of capillaries or skin response to cold temperatures; as an additional

prevention measure, participants washed with hypoallergenic goat’s milk liquid pump soap out of

consideration for both hygiene and minimising the potential for an allergic reaction. Once washed,

participants would then dry their hands with paper towel under instruction to ensure that their

palms did not remain damp. Participants were then guided back to their chair, whereupon the

attachment of all psychophysiological measures occurred. Two disposable EL507 snap electrodes

were attached to the thenar and hypothenar regions of the palm (see Figure 32) respectively, and

secured with medical tape to reduce the risk of movement or detachment throughout the

experiment. The real-time EDA recording within the AcqKnowledge recording and analysis

software was then visually verified to ensure connection and an uninterrupted EDA trace. EDA

was attached first to allow for the 10-minute ‘settle’ time to occur throughout the attachment of

the additional psychophysiological measures.

The GSR100C Bioamp module settings for EDA recording were as follows:

GAIN: 5µΩ/V

LP: 10Hz

HP: DC

HP: DC.

137


5.1.12 EMG Attachment

Participants’ skin was topically abraded in the left OO, left CS and grounding regions; to

ensure that each muscle was accurately located, participants were asked to smile (OO) and frown

(CS) while the experimenter held a cotton bud against the target site. Once the muscle was located,

the skin was abraded for approximately 60 seconds with Nuprep Skin Prep Gel applied to a cotton

bud to strip the top layer of skin particles. The gel was then thoroughly wiped away with gauze

pad, doubling as a secondary abrasion process. The abraded sites were further cleaned with an

alcohol wipe, removing any remaining product or skin debris.

Prior to the participant’s arrival, adhesive collars were attached to four shielded EL254s

4mm Ag-AgCl cup electrodes (for use in measuring EMG OO and EMG CS) and a single

unshielded EL254 4mm Ag-AgCl cup electrode (for employment as a grounding electrode). Once

the EMG skin regions had been prepared, the electrode cup cavities were filled with

hypoallergenic Signa conductive gel; this process was delayed until immediately prior to

attachment to prevent the gel drying. The single unshielded EL254 ground electrode was then

fixed to the centre of the forehead; similarly, a pair each of the EL254S shielded electrodes were

next attached to the OO and CS regions (see Figure 28). Electrode leads were tucked behind the

participant’s left ear to prevent vision obstruction. All electrodes were secured with medical tape.

Finally, impedance between both the OO electrode pair and CS electrode pair was checked using

the UFI Model 1089 MkIII Checktrode; in the event of unsatisfactory impedance levels (> 10kΩ),

the electrodes were removed and the skin preparation and application process restarted with new

electrodes.

138


The electrode leads fed into an extension module that was clipped to the participant’s shirt

collar, as the original lead length for the electrodes was too restrictive of movement. As with EDA,

the EMG recordings were checked against AcqKnowledge to ensure an uninterrupted connection.

The EMG100C Bioamp module settings for EMG recording were as follows:

GAIN: 2000

LP: 500 Hz

HP: OFF

HP: 1.0 Hz.

5.1.13 Electrocardiography Attachment

The electrode sites for the two-lead ECG recording followed the same abrasion procedure

outlined above in section 5.1.7.2. The sites chosen were approximately three to five centimetres

below the right clavicle, and on the lower left-hand side of the ribcage (approximately three

centimetres above the lowest ribcage bone, or roughly in alignment with the elbow); see Figure

29. All participants wore a two-piece outfit (e.g., trousers and a shirt) to the experiment, in

Figure 30. EMG placement. EMG CS electrodes placed on brow, EMG OO electrodes placed near eye.

Right: Ground electrode visible in centre of forehead.

139


compliance with a request given in the recruitment materials, to ensure appropriate access to these

regions.

As with the EMG, as outlined in the previous section, EL254s 4mm Ag-AgCl electrodes

were prepared, but unfilled, prior to the participant’s arrival to avert complications that could arise

with dried electrode gel. Once the sites had been prepared, the electrode gel was added to the cup

cavities and attached to the participant. Both electrodes were then secured with medical tape and

fed into an extension module clipped to the participant’s shirt collar. Impedance was then checked

on the UFI Model 1089 MkIII Checktrode, with impedance levels below 10kΩ necessitating

removing the electrodes and restarting the preparation and application process.

As with EMG and EDA, recordings were checked against AcqKnowledge to ensure an

uninterrupted connection. In some cases, large amounts of adipose tissue prevented the obtainment

of an adequate signal from the ECG electrodes. If this issue continued after removal and

reapplication of new electrodes, the experiment proceeded in the interests of time restraints, with

ECG data marked for later removal from the sample.

Figure 31. ECG placement.

140


The ECG100C Bioamp module settings for ECG recording were as follows:

GAIN: 500

MODE: NORM

LPN: ON

HP: 0.5Hz.

5.1.14 Electroencephalography Attachment

Prior to the participant’s arrival at the laboratory, the EEG headset’s Epoc 14 felt sensor

pads were moistened with a saline solution. Additionally, between experiment sessions, the sensor

pads were kept in a hydrator pack to minimise drying. Immediately prior to attaching the EEG

headset, the sensor pads were again briefly remoistened. Finally, in the continued absence of a

stable or present connection with specific electrode sites, the participant’s scalp was topically

doused with the saline solution.

Once attached to the participant’s scalp, the EEG headset was adjusted to ensure adequate

contact with all electrode regions. This information was provided by the EMOTIV Epoc

TestBench recording software: if connection was established, the circle representing the relevant

electrode site remained blank. Colour signified contact quality in all other instances: red, orange,

yellow and green circles indicated contact quality in ascending order (red indicated the lowest

contact quality and green indicated the highest). Data collection only occurred once all electrode

sites displayed as green. See Figures 30 and 31 for the contact quality display and headset

placement.

141


5.1.15 Data Treatment

For this program of research, physiological data was evaluated tonically over the course

of each play session. A tonic analysis approach was selected for analysis to allow for an overview

of the player experience. Phasic analysis thus represents an additional path for analysis in future

research. Tonic analysis was also found to be congruent with this program’s research questions

concerning evaluation of utility in employing psychophysiological measures to predict subjective

states that reflect the entirety of the play experience. Further discussion on this approach is detailed

in sections 6.4 and 6.5.

All physiological data was acquired on two programs installed on a machine removed

from the participant’s field of view. Recording times were manually flagged and timestamped

within both programs by the researcher as they occurred (e.g., when a play session began, when a

play session terminated, when a baseline began and so on). Although each play session was 10

minutes and 30 seconds in length, only 10 minutes of each play session was analysed; 20 seconds

were removed from the start of the play session and 10 seconds removed from the conclusion.

Figure 35. TestBench contact

quality display.

Figure 34. EEG placement.

142


This was to allow participants a grace period to physically settle or adjust in the first 20 seconds

of play, and to allow the researcher a delay when flagging the end of the play session during the

last 10 seconds.

All physiological data underwent a rigorous cleaning process prior to its inclusion for

statistical analysis, as detailed in the following sections. These also describe the treatment and

analysis methods used per psychophysiological measure.

5.1.16 Electrodermal Activity

EDA activity (mS) was recorded in BIOPAC’s AcqKnowledge 4.2 data acquisition and

analysis software. Once collected, the data were manually inspected for movement artefacts, noise

interference and interrupted signals. From the full sample size of 89, four cases were removed

from analysis due to largely compromised or non-existent data. It is speculated that this loss of

data occurred as a consequence of an electrode shifting or detaching from the site; in particular,

this may have transpired due to sweat eroding the adhesion of both the electrode and the tape. In

all cases, movement artefacts were visually evaluated in windows (timebins) of 10-second epochs.

If an artefact was found to influence two or more seconds of the epoch, the full 10-second epoch

was removed from analysis. As with all measures, any cases that contained over 15% of data loss

were removed from analysis.

The values corresponding with the 10-minute play sessions, with movement and noise

artefacts handled through null value replacement, were then exported into SPSS for analysis. A

total mean value of each 10-minute play session was derived per n.

143


5.1.17 EMG OO and CS

EMG activity (µV) was recorded in BIOPAC’s AcqKnowledge 4.2 data acquisition and

analysis software. Once collected, the data was manually inspected for movement artefacts, noise

interference and interrupted signals. For EMG OO, 31 cases were removed from analysis due to

largely compromised or non-existent data; for EMG CS, 54 cases were removed from analysis. In

all instances of removed cases, the EMG trace indicated either a loss of signal or a signal

compromised by an excess of noise (see Figure 33). This may have occurred either as a

consequence of lost contact between the electrode and electrode site during data collection, poor

electrode contact or faulty equipment. In instances where the contaminated data was minor, visual

inspection and handling of artefacts occurred as in the treatment of EDA data (see section 5.1.16).

For the analysis and treatment of EMG data, a 10 Hz high-pass filter using a Hanning

window was first applied to the entire data channel; this was undertaken for the purpose of

clarifying the EMG data through filtering out of most eye movement and blink artefacts. The

Figure 36. EDA placement.

144


average rectified EMG was then derived from each channel individually at intervals of 0.5

seconds. A mean epoch analysis, using 10-second epoch widths with zero time lapse between

epochs, was performed on the rectified data. The values corresponding with the 10-minute play

sessions, with movement and noise artefacts handled through null value replacement, were then

exported into SPSS for analysis.

5.1.18 Electrocardiography

ECG activity was recorded in BIOPAC’s AcqKnowledge 4.2 data acquisition and analysis

software. Once collected, the data was manually inspected for movement artefacts, noise

interference and interrupted signals. Of the full sample size of 89, nine cases were removed from

analysis due to compromised or non-existent data. As with EDA, it is speculated that this loss of

data occurred as a consequence of an electrode shifting or detaching from the site. In some cases,

a consistent ECG signal was unobtainable with the available electrodes due to excess adipose

tissue in the lower abdominal region.

Figure 38. EMG data comparison. Top left and bottom indicate noisy or weak signals,

respectively; top right indicates usable/uncompromised data. Top images both indicate

single 10-second epochs, bottom image indicates 2 x 15-second epochs.

145


Once visual inspection of the data was completed, the 10-minute time bins corresponding

with play sessions were separated and imported into Kubios HRV 2.2 HR variability analysis

software (Tarvainen, Niskanen, Lipponen, Ranta-oho, & Karjalainen, 2014). Mean R-R, HR, HF

peaks (Hz), and LF/HF analyses were performed (HR and HF peaks were evaluated within this

program of research). The window width used for the fast Fourier transform was 256 samples,

with a window overlap of 50%. The frequency bands used for frequency analysis were 0.04–0.15

for LF and 0.15–0.4 for HF. All analyses were undertaken using the Kubios software’s native

artefact correction filter, set to the level ‘Very Strong’, which allowed for the correction of out-

of-range R-R intervals and featured respiration frequency analysis that ensured the HF component

remained within the HF band limits (Tarvainen et al., 2014). The final values for HF peaks (Hz)

and HR were exported to SPSS for statistical analysis.

5.1.19 EEG

EEG activity (uV) was recorded in EMOTIV’s TestBench data acquisition software, with

BIOPAC’s AcqKnowledge 4.2 software employed for data treatment and frequency analysis. For

the purposes of this research program, the AF4 and O2 sites were evaluated for their proximity to

the frontal and occipital regions (see section 3.3.3.1). Absolute power for the frequency bands

were derived within SPSS for the AF4 and O2 sites respectively, with average power estimates

calculated using a fast Fourier transformation within AcqKnowledge’s frequency analysis. The

thresholds used to identify each frequency band were as follows:

Theta: 4–8 Hz

Alpha: 8–13 Hz

Beta: 13–30 Hz.

146


Upon frequency analysis output, each frequency band was visually inspected for

movement artefacts, noise interference and interrupted signals. Of the full sample of 89, 22 cases

were lost for the AF4 site and 20 cases were lost for the O2 site. As with EDA and ECG, it is

speculated that a proportion of this data loss occurred as a consequence of electrodes shifting or

detaching from the site; however, in some cases, regular interference throughout rendered much

of the signal unusable. For cases with minor movement artefacts and interference, the severity and

length of the artefact was manually assessed and replaced with null values when appropriate.

The mean absolute power values for each play condition per frequency band per electrode

site were then imported into SPSS. A mean value was then derived from each of these for statistical

analysis, producing a total of 18 separate EEG variables per participant (e.g., O2 Alpha for

Boredom, O2 Alpha for Balance, O2 Alpha for Overload and so on, for alpha, beta and theta across

both sites).

5.1.20 Participants

Ninety participants participated in the study, but a single participant was removed from

the final sample due to computer hardware failure prematurely terminating the experiment. Of the

final sample size of 89 participants, 77.5% were male and 22.5% were female, all between the

ages of 17 and 38 (mean = 23.41, SD = 4.53). On a Likert scale of 1–7, with ‘7’ representing

‘extremely experienced’ and ‘1’ ‘not at all experienced’, participants self-rated with a mean of

5.96 (SD = 1.33) for ‘general experience with video games’ and 5.16 (SD = 1.72) for ‘experience

with first-person shooters’. Participants reported between 0 and 70 hours spent playing video

games per week, with a mean of 21.79 hours (median = 20 hours, SD = 16.04). In terms of

familiarity with the Left 4 Dead franchise, participants were asked how many hours they had spent

playing both Left 4 Dead 1 and Left 4 Dead 2. Forty-five participants had never played Left 4 Dead

147


1 before, with the remaining 44 having played between 1 and 120 hours (mean = 25.61 hours,

median = 10 hours, SD = 36.48). As for Left 4 Dead 2, 34 had never played before, and the

remaining 55 had played between 1 and 200 hours (mean = 32.47, median = 15 hours, SD = 43.43).

Overall, 69.66% of participants had previously played a Left 4 Dead title.

5.1.21 Analysis

To explore RQ1a, the subjective and psychophysiological response to the Boredom,

Balance and Overload conditions were evaluated. Subjective player experience measures and

psychophysiological measures were analysed separately. The primary analysis was undertaken

using a one-way MANOVA to determine the effect of condition (Boredom, Balance, Overload)

on subjective player experience (flow, enjoyment, competence, autonomy and presence) and a

combination of one-way MANOVAs and ANOVAs for the psychophysiological measures.

Specifically, the majority of the psychophysiological measures (EDA, HR, HF; all EEG

frequencies) were assessed together in a single MANOVA, while EMG CS and EMG OO were

each assessed in individual ANOVAs due to a notably diminished sample size as a consequence

of data loss. The exploration of subjective response was undertaken as a manipulation check to

evaluate the conditions’ respective success in either evoking or inhibiting an optimal play

experience. The MANOVA approach offers an advantage in psychophysiological analysis in that

it does not make the assumption of sphericity, which is rarely met in psychophysiological data

(Vasey & Thayer, 1987).

To explore how well psychophysiological response predicted subjective states (RQ2), two

sets of multiple regression analyses were initially conducted. The first set focused on the

participant’s psychophysiological response and player experience ratings in the Balance condition

(one regression equation was calculated for each subjective player experience construct, with

148


psychophysiological measures entered as predictor variables). To allow for the possibility that

some participants may have experienced play more positively (e.g., greater flow) in a condition

other than Balance, a second set of regression analyses was conducted using each participant’s

data from the condition in which they experienced the highest levels of the outcome measure (e.g.,

flow, interest/enjoyment, etc.). Based on the results of the regression analyses, further data

exploration was undertaken using Pearson’s correlation coefficients and calculations of

variability. All statistical analyses were undertaken using IBM SPSS Statistics 23.0.

5.2 SELF-REPORT RESULTS

5.2.1 Confirmation of Optimal and Sub-optimal Conditions

5.2.1.1 Assumptions and Outliers for Subjective Measures

Mahalanobis distance identified one multivariate outlier, and boxplots revealed two

univariate outliers for flow (Boredom, Overload) and four univariate outliers for

interest/enjoyment (Boredom) (Field, 2013). No substantive differences in the pattern of results

were found with outliers excluded from analyses; in the interest of statistical power, the results

reported here include all 89 cases.

Preliminary assumption-checking revealed abnormalities in data distribution, as assessed

by the Shapiro-Wilk test (p > .05). The violations of normality occurred for the subjective ratings

of flow in the Boredom (p = .036) and Balance (p = .014) conditions; ratings of interest/enjoyment

in the Boredom condition (p = < .001); ratings of competence in the Boredom (p = .006), Balance

(p = < .001) and Overload (p = .001) conditions; ratings of autonomy in the Boredom condition

(p = .045); and ratings of presence in the Boredom (p = .012) and Balance (p = .048) conditions.

The analyses were subsequently run on both untransformed and transformed data; however (as

149


might be expected given the robustness of the one-way MANOVA to deviations from normality

[Laerd, 2013]), the pattern of results remained the same, so for ease of interpretation, the

untransformed results are reported in this section. There was no evidence of multicollinearity, as

assessed by Pearson correlation; there were linear relationships, as assessed by scatterplots.

5.2.1.2 Findings

A repeated-measures MANOVA revealed a significant multivariate within-subjects effect

of condition on flow, interest/enjoyment, competence, autonomy and presence using Wilk’s

Lambda (F(10, 79) = 26.807, p < .001, ηp2 = .772). Mauchly’s Test indicated that the assumption

of sphericity had been violated for flow (W = .870, χ2(2) = 12.100, p < .005), interest/enjoyment

(W = .661, χ2(2)= 35.989, p < .001), competence (W = .887, χ2(2)= 10.417, p = .005) and autonomy

(W = .837, χ2(2) = 15.505, p < .001), and so a Greenhouse-Geisser adjustment (flow: ε = .885;

interest/enjoyment: ε = .747; competence: ε = .899; autonomy: ε = .860) was used for these DVs

in within-subjects univariate analysis. Sphericity was assumed for presence. Significant univariate

main effects were observed for flow (F(1.770, 155.773) = 34.634, p < .001, ηp2 = .282),

interest/enjoyment (F(1.494, 131.463) = 33.693, p < .001, ηp2 = .277), Competence (F(1.797,

158.153) = 85.387, p < .001, ηp2 = .492), autonomy (F(1.719, 151.301) = 27.332, p < .001, ηp

2 =

.237) and presence (F(2, 176) = 29.529, p < .001, ηp2 = .251).

Post-hoc analysis revealed that for the main effect on flow, the Balance condition (M =

3.92, SD = 0.553) showed significantly higher flow levels than the Overload condition (M = 3.447,

SD = 0.588, p < .001); likewise, the Boredom condition (M = 3.931, SD = 0.472) showed

significantly higher flow levels than the Overload condition (p < .001). No significant differences

were revealed between the Balance and Boredom conditions (p > 0.999). For the main effect on

interest/enjoyment, participants reported significant differences between all three conditions, such

150


that the Balance condition (M = 4.728, SD = 1.146) showed significantly greater

interest/enjoyment than both the Boredom (M = 3.47, SD = 1.185, p < .001) and Overload (M =

4.233, SD = 1.35, p < .001) conditions, and the Overload condition also showed higher

interest/enjoyment than the Boredom condition (p < .001). For the main effect on competence,

both Boredom (M = 5.045, SD = 1.374, p < .001) and Balance (M = 4.838, SD = 1.489, p < .001)

showed higher levels of competence than Overload (M = 2.857, SD = 1.342), with no significant

differences revealed between Boredom and Balance (p = .842). For the main effect on autonomy,

the Balance condition (M = 4.247, SD = 1.316) showed significantly higher levels of autonomy

than both the Boredom (M = 3.299, SD = 1.42, p < .001) and Overload (M = 3.326, SD = 1.253, p

< .001) conditions, with no significant differences revealed between the Boredom and Overload

conditions (p > 0.999). Finally, the main effect on presence revealed significantly higher levels of

presence in the Balance condition (M = 3.88, SD = 1.34) than in both the Boredom (M = 3.054,

SD = 1.231, p < .001) and Overload (M = 3.298, SD = 1.252, p < .001) conditions, with no

significant difference revealed between Boredom and Overload conditions (p = .104). See Table

5 for a summary of these results, and Figures 34‒38 for visualisations of these results.

151


Table 5. Summary of main effect on subjective response

Sample Size Mean SD Bore p Bal p Over

p

Flow n = 89

Boredom 3.931 0.472 < .001

Balance 3.92 0.553 < .001

Overload 3.447 0.588 < .001 < .001

Int/Enj n = 89

Boredom 3.47 1.185 < .001 < .001

Balance 4.728 1.146 < .001 < .001

Overload 4.233 1.35 < .001 < .001

Competence n = 89

Boredom 5.045 1.374 < .001

Balance 4.838 1.489 < .001

Overload 2.857 1.342 < .001 < .001

Autonomy n = 89

Boredom 3.299 1.42 < .001

Balance 4.247 1.316 < .001 < .001

Overload 3.326 1.253 < .001

Presence n = 89

Boredom 3.054 1.231 < .001

Balance 3.88 1.34 < .001 < .001

Overload 3.298 1.252 < .001

152


Figure 39. Study 2 Short Flow State Scale results.

Figure 40. IMI Interest/Enjoyment results.

153


Figure 42. PENS presence results.

Figure 41. PENS competence results.

Figure 43. PENS autonomy results.

154


5.3 PSYCHOPHYSIOLOGICAL DIFFERENCES IN OPTIMAL AND SUB-

OPTIMAL CONDITIONS

5.3.1 Assumptions and Outliers for Psychophysiological Measures

There were no multivariate outliers as assessed by Mahalanobis distance. However, box

plot assessment revealed univariate outliers across conditions: three unique outliers were

identified for EDA, two for HR, five for HF, seven for EMG OO and three for EMG CS.

Additionally, outliers were found at many of the EEG collection sites for one or more of the

frequency bands; the number of outliers ranged from to four to 10 (for details, see Appendix F).

No substantive differences in the pattern of results were found with outliers for EDA, HR, HF and

EMG CS; analysis was thus performed on all cases. As outliers were found to be influencing the

pattern of results for EMG OO and EEG, all analyses were performed on transformed versions of

this data. As outliers remained in the transformed EEG data, a further four outliers identified as

consistent across all sites and frequency bands were removed from analysis.

Preliminary assumption-checking revealed abnormalities in data distribution, as assessed

by the Shapiro-Wilk test (p > .05). The violations of normality occurred for measures of HF in all

conditions (p < .001); EMG CS in the Boredom (p = .028) and Balance (p = .039) conditions;

transformed EMG OO in the Balance condition (p = .005); AF4 Alpha in the Balance condition

(p = .034); and AF4 Theta in the Overload condition (p = .012). The analyses were subsequently

run on both untransformed and transformed data for HF and EMG CS; however, the pattern of

results remained the same, so for ease of interpretation, the untransformed results are reported in

this section. These are further supported by the robustness of the one-way MANOVA to deviations

from normality (Laerd, 2013). There was no multicollinearity, as assessed by Pearson correlation.

Some evidence of non-linearity was identified for EMG OO, HF peaks and all EEG frequency

155


bands across both sites, suggesting that analyses involving these variables had reduced statistical

power (Laerd, 2013).

Due to disparities in sample sizes between measures, as discussed in sections 5.1.8.2,

5.1.8.4 and 5.1.10, analyses of these measures were conducted in a series of separate one-way

MANOVAs and ANOVAs. Analyses were thus performed on EDA, HR and HF in a single one-

way MANOVA on 75 cases; all EEG sites and frequencies in a separate single one-way

MANOVA on 50 cases; and EMG OO and EMG CS in individual one-way RM ANOVAs, with

58 and 35 cases respectively.

5.3.2 Results for the One-Way Multivariate Analysis of Variance on

Electrodermal Activity, High Frequency and Heart Rate


of condition on EDA, HF and HR using Wilk’s Lambda (F(6, 69) = 9.041, p < .001, ηp2 = .440).

Mauchly’s Test indicated that the assumption of sphericity had been violated for HR (W = .891,

χ2(2) = 8.399, p = .015), and so a Greenhouse-Geisser adjustment (ε = .902) was used for this DV

in within-subjects univariate analysis. Sphericity was assumed for EDA and HF. Significant

univariate main effects were observed for EDA (F(2, 148) = 6.538, p = .002, ηp2 = .081), HR

(F(1.804, 131.682) = 5.632, p = .006, ηp2 = .071) and HF (F(2, 148) = 8.396, p < .001, ηp

2 = .102).

Post-hoc analysis revealed that for the main effect on EDA, the Overload condition (M =

15.101, SD = 6.65) showed significantly higher EDA levels than the Boredom condition (M =

14.107, SD = 6.616, p = .001); no significant differences were revealed between Overload and

Balance conditions (M = 14.689, SD = 6.61, p = .540) or the Balance and Boredom conditions (p

= .095). For the main effect on HR, analysis revealed significantly greater HR in the Boredom

condition than in the Overload condition, such that the Boredom condition (M = 79.075, SD =

156


13.284) showed significantly higher HR than the Overload condition (M = 76.571, SD = 13.564,

p < .001); no significant differences were revealed between the Boredom and Balance conditions

(M = 78.046, SD = 13.194, p = .684) or the Balance and Overload conditions (p = .166). Finally,

for the main effect on HF Peaks, the Overload condition (M = .208, SD = .0733) revealed

significantly lower HF Peaks than both the Balance (M = .236, SD = .086, p = .018) and Boredom

(M = .248, SD = .078, p < .001) conditions; no significant difference was found between Balance

and Boredom (p = .760). For a summary of these results, see Table 6; for a visualisation, see

Figures 39‒41.

Table 6. Summary of main effect on EDA, HR, and HF Peaks

Sample Size Mean SD Bore p Bal p Over

p

EDA n = 75

Boredom 14.107 6.616 = .001

Balance 14.689 6.61

Overload 15.101 6.65 = .001

Heart Rate n = 75

Boredom 79.075 13.284 < .001

Balance 78.046 13.194

Overload 76.571 13.564 < .001

HF Peaks n = 75

Boredom .248 .078 < .001

Balance .236 .086 = .018

Overload .208 .0733 < .001 = .018

157


Figure 44. EDA results

Figure 45. HRV HF Peaks results

Figure 46. HR results

158


5.3.3 Results for the One-Way Multivariate Analysis of Variance on

Electroencephalography


of condition on all transformed EEG variables using Wilk’s Lambda (F(12, 38) = 35.850, p < .001,

ηp2 = .919). Mauchly’s Test indicated that the assumption of sphericity had been violated for AF4

Alpha (W = .833, χ2(2) = 8.767, p = .012), AF4 Beta (W = .827, χ2(2) = 9.126, p = .010), O2 Alpha

(W = .364, χ2(2) = 48.556, p < .001), O2 Beta (W = .246, χ2(2) = 67.318, p < .001) and O2 Theta

(W = .334, χ2(2) = 52.617, p < .001), and so a Greenhouse-Geisser adjustment (AF4 Alpha: ε =

.857; AF4 Beta: ε = .852; O2 Alpha: ε = .611; O2 Beta: ε = .570, O2 Theta: ε = .600) was used for

these DVs in within-subjects univariate analysis. Sphericity was assumed for AF4 Theta.

Significant univariate main effects were observed for AF4 Beta (F(1.705, 83.536) = 7.793, p =

.001, ηp2 = .137), O2 Alpha (F(1.222, 59.889) = 24.767, p < .001, ηp

2 = .336), O2 Beta (F(1.140,

55.872) = 31.171, p < .001, ηp2 = .389) and O2 Theta (F(1.201, 58.829) = 28.333, p < .001, ηp

2 =

.366).

Post-hoc analysis revealed that for the main effect on transformed AF4 Beta, the Boredom

condition (M = -2.474, SD = .163) showed significantly lower AF4 Beta than both the Balance (M

= -2.418, SD = .189, p = .008) and Overload (M = -2.428, SD = .174, p = .012) conditions; no

significant differences were found between the Balance and Overload conditions (p > .999). For

the main effect of transformed O2 Alpha, the Boredom condition (M = -2.608, SD = .185) showed

significantly lower O2 Alpha than both the Balance (M = -2.547, SD = .183, p < .001) and

Overload (M = -2.571, SD = .169, p < .001) conditions; no significant differences were found

between the Balance and Overload conditions (p > .999). For the main effect of transformed O2

Beta, the Boredom condition (M = -2.491, SD = .218) showed significantly lower O2 Beta than

159


both the Balance (M = -2.404, SD = .232, p < .001) and Overload (M = -2.443, SD = .205, p <

.001) conditions; no significant differences were found between the Balance and Overload

conditions (p = .261). Finally, for transformed O2 Theta, the Boredom condition (M = -2.506, SD

= .182) showed significantly lower O2 Theta than both the Balance (M = ‒2.448, SD = .169, p <

.001) and Overload (M = ‒2.465, SD = .163, p < .001) conditions; no significant differences were

found between the Balance and Overload conditions (p > .999). No significant differences between

conditions were found for the main effects on the AF4 Alpha and AF4 Theta DVs. See Figure 42

for a visualisation of these results.

5.3.4 Results for the One-Way Repeated-Measures Multivariate Analysis of

Variance on Electromyography - Orbicularis Oculi

As discussed in sections 5.1.8.2 and 5.1.10, EMG OO was excluded from the RM

MANOVA analysis of psychophysiological variables due to the disparity in sample size.

Therefore, in an effort to preserve statistical power for EDA, HR and HF, the transformed data for

EMG OO was assessed in a separate, one-way repeated-measures ANOVA.

Figure 47. EEG frequency band results (reversed).

160


Mauchly’s Test indicated that the assumption of sphericity had been violated (W = .810,

χ2(2) = 11.820, p = .003), and so a Greenhouse-Geisser adjustment (ε = .840) was used for analysis.

A one-way repeated-measures ANOVA revealed a significant within-subjects effect of condition

on EMG OO (F(1.680, 95.775) = 22.007, p < .001, ηp2 = .279).

Post-hoc analysis revealed that for the main effect on transformed EMG OO, the Overload

condition (M = -2.33, SD = .246) showed significantly higher levels of EMG OO than either the

Balance (M = -2.412, SD = .229, p = .001) or Boredom (M = -2.462, SD = .222, p < .001)

conditions. Additionally, the Balance condition showed significantly higher levels of EMG OO

than the Boredom condition (p = .005). See Figure 43 for a visualisation of these results.

Figure 48. EMG OO results (reversed).

161


5.3.5 Results for the One-Way Repeated-Measures Multivariate Analysis of

Variance on Electromyography - Corrugator Supercilii

As with EMG OO, EMG CS was excluded from the RM MANOVA analysis of

psychophysiological variables due to disparities in sample size. Therefore, in an effort to preserve

statistical power for EDA, HR and HF, the data for EMG CS was assessed in a separate one-way

repeated-measures ANOVA.

Mauchly’s Test indicated that the assumption of sphericity had been violated (W = .678,

χ2(2) = 12.817, p = .002), and so a Greenhouse-Geisser adjustment (ε = .757) was used for analysis.

A one-way repeated-measures ANOVA revealed no significant within-subjects effects of

condition on EMG CS (F(1.531, 51.443) = 2.492, p = .106, ηp2 = .068).

5.3.6 Results for Exploration of Predictive Relationships

To assess RQ2, a series of multiple regressions employing the psychophysiological

measures as the predictors and each of the subjective player experience measure as the outcome

measure (e.g., predicting the flow ratings from EDA, EMG OO, HR, HF and EEG) was

undertaken. The initial set of regression analyses was conducted on data from the Balance

condition, where evaluation of the subjective reports had established that the most optimal player

experience was occurring. After assumption-checking, all regression equations were found to be

non-significant (p > .05). Following this, the regression analysis was conducted using data for

each participant drawn from the condition in which they experienced the highest levels of the

outcome measure. For example, for the regression equation predicting interest/enjoyment,

participants were filtered by highest rating of interest/enjoyment; predictions were then attempted

on that value from the psychophysiological data obtained in the condition in which they

162


experienced the greatest interest/enjoyment. Again, after assumption-checking, all regression

analyses were found to be non-significant (p > .05).

To further analyse the data, correlations were calculated between the subjective and

physiological measures. Very few significant correlations were found (see Table 7 for the Balance

condition correlations). As a final further step to try to understand the cause of this lack of

correlation, variances were calculated for each variable; the results show that many of the variables

had low variance (see Table 8).

163


Table 7. Correlations between subjective and psychophysiological measures

Flow Int/En

j

Comp Auto Pres

EDA Pearson Correlation .053 .033 ‒.032 ‒.048 ‒.094

n = 81 Sig. (2-tailed) .635 .767 .774 .668 .403

EMG OO Pearson Correlation .110 .119 .223 .406 .299

n = 54 Sig. (2-tailed) .428 .392 .105 .002 .028

EMG CS Pearson Correlation .258 .253 .354 .055 ‒.011

n = 33 Sig. (2-tailed) .147 .155 .043 .761 .953

HR Pearson Correlation ‒.140 .035 ‒.205 ‒.026 -.006

n = 71 Sig. (2-tailed) .243 .774 .087 .828 .963

HF Peak Pearson Correlation ‒.052 ‒.059 ‒.139 .030 .099

n = 71 Sig. (2-tailed) .669 .627 .246 .807 .409

AF4 Alpha Pearson Correlation ‒.106 .071 ‒.014 ‒.022 .005

n = 60 Sig. (2-tailed) .420 .588 .916 .865 .971

AF4 Beta Pearson Correlation ‒.097 .004 .011 .036 .070

n = 58 Sig. (2-tailed) .469 .974 .932 .788 .603

AF4 Theta Pearson Correlation ‒.145 ‒.059 ‒.091 -.010 .013

n = 59 Sig. (2-tailed) .274 .657 .493 .940 .923

O2 Alpha Pearson Correlation ‒.115 .149 .039 .076 .093

n = 61 Sig. (2-tailed) .376 .251 .767 .560 .475

O2 Beta Pearson Correlation ‒.109 .122 .033 .087 .010

n = 59 Sig. (2-tailed) .413 .357 .807 .513 .943

O2 Theta Pearson Correlation ‒.138 .061 ‒.049 .018 .154

n = 60 Sig. (2-tailed) .291 .643 .709 .893 .241

164


Table 8. Variances for all variables within conditions

Psychophysiological

Measures


EDA 45.571 47.154 46.485

ECG—HR 175.344 176.744 190.608

ECG—HF Peaks 0.006 0.007 0.005

EMG OO (below eye) 0.051 0.055 0.062

EMG CS (brow) - - -

EEG AF4 Alpha 0.034 0.034 0.03

EEG AF4 Beta 0.04 0.038 0.037

EEG AF4 Theta 0.041 0.034 0.036

EEG O2 Alpha 0.037 0.04 0.037

EEG O2 Beta 0.057 0.058 0.053

EEG O2 Theta 0.031 0.03 0.033

Subjective Measures Boredom Balance Overload

Flow 0.199 0.319 0.348

IMI Enjoyment 1.46 1.362 1.878

PENS Autonomy 2.064 1.754 1.64

PENS Competence 1.859 2.297 1.869

PENS Presence 1.497 1.779 1.611

165


5.4 DISCUSSION

5.4.1 Subjective Experience of Play

The play conditions developed for this study follow precedents established in prior

psychophysiological research in the player experience space (Nacke, 2008; Keller & Bless, 2008;

Keller et al., 2011). The conditions were designed to compare optimal and sub-optimal play

experiences in a quantifiable way, as in research undertaken by Nacke et al. (2008), Keller and

Bless (2008) and Keller et al. (2011). To achieve this, the manipulation of challenge‒skill

balance—an approach also employed by Keller and Bless—was selected for the condition design.

While challenge‒skill balance was originally selected for its integral role played in achieving flow

(Csikszentmihalyi, 1990), the manipulation remained relevant to the research program despite

widening the focus to include additional psychological constructs. The use of three manipulations

designed to support or inhibit the optimal play experience (challenge‒skill balance, challenge

outstripping skill and skill outstripping challenge) provided a means of comparing psychological

and psychophysiological responses.

5.4.2 IMI Interest/Enjoyment

As expected, results for the interest/enjoyment subscale of IMI showed significantly

greater participant interest/enjoyment in the Balance condition than in the Boredom and Overload

conditions. Separately, participants also reported greater interest/enjoyment in the Overload

condition than in the Boredom condition. These results thus support the Balance condition—in

relation to the Overload and Boredom conditions—as the optimal play experience in terms of

interest/enjoyment. These findings also offer further evidence that challenge‒skill balance plays

an important role in enabling a positive play experience, as supported by other research (Keller

and Bless, 2008; Nacke and Lindley, 2008).

166


Interestingly, the significant player preference for Overload over Boredom in terms of

interest/enjoyment has implications for assessing play experience that contains challenge‒skill

imbalances. This may tentatively point to challenge‒skill imbalance, wherein challenge outstrips

skill, as a more enjoyable experience than the reverse—that is, overwhelming the player may lead

to more positive experiences or enjoyment than underwhelming or boring the player. This is

further strengthened by research undertaken by Abuhamdeh and Csikszentmihalyi (2009), in

which levels of enjoyment among chess players were at their highest when players were up against

better opponents, as opposed to those of equal ability; when perceived challenges were higher than

player skill, the games were rated as more enjoyable. While these conclusions are not wholly

applicable to the pattern of results in this study, in which the Balance condition was rated as more

enjoyable than the Overload condition, this may assist in explaining the disparities between

enjoyment of Boredom and Overload.

Another explanation may be that the presence of combat in a combat-based game,

regardless of challenge‒skill balance or imbalance, is inherently more fun than or preferable to

the complete absence of combat. As such, replication of this study with other game genres may be

useful in future iterations of this work. However, when considering previous research that suggests

a preference for low challenge among casual or inexperienced players (Alexander, Sear, &

Oikonomou, 2013; Lomas, Patel, Forlizzi, & Koedinger, 2013), it is important to consider that the

findings of the present study may reflect the relatively high levels of self-reported experience

among participants.

It is also possible that the Overload condition remained ‘fun’ for some individuals despite

the excessive challenge of the condition. While the monotony of the Boredom condition ensured

a consistent experience throughout play, the Overload condition may have allowed opportunities

167


for moments of enjoyment—for example, when the participants were not yet aware of the

condition’s impossibility, or as a consequence of any minor progress in the face of overwhelming

odds. This was corroborated anecdotally through comments made by participants in a ‘General

Comments’ field in the respective condition surveys, including the following:

‘I had more determination to complete the objective even though I kept dying. Just getting

one or two cans was enough.’ (P210)

‘It had its moments, such as when everyone was away from me and I was healing up

myself, only at the VERY LAST MOMENT a zombie hitting me in the back—that was

kind of cool.’ (P113)

'I found it both more enjoyable and entertaining when being greatly challenged and failing

than when not challenged at all.’ (P132)

5.4.3 Flow

Flow findings generally followed the same pattern of results discovered in Chapter 4;

while both Boredom and Balance induced significantly greater levels of flow than Overload, no

differences between Boredom and Balance were revealed. This may point to the same possible

conclusions established in the pilot study discussion:

that the immersive quality of highly detailed environments may still contribute to

experiences of flow, regardless of the challenge of the experience

that the condition design was unsuccessful in disinhibiting flow, or that challenge may

still be found in unexpected ways

that challenge‒skill balance may not be a strong antecedent or precursor to flow, despite

common conception

168


or finally, that Jackson’s flow state scale (in this case, the S FSS-2) may not be applicable

as a measure of flow in play experiences.

As the original pilot study results directly informed the redesign of the Boredom

condition, these conclusions may be expanded. Removing combat from the Boredom condition

minimised the challenge of the game as much as possible while still retaining game-like elements

(primarily in the collection and retrieval of the gas canisters scattered throughout the map). It thus

seems unlikely that any unexpected challenge was a consequence of limited or improper condition

design. The success of the condition design in achieving boredom through challenge‒skill

imbalance was corroborated anecdotally by feedback provided in the condition survey responses:

‘Seeing I was just walking around the game space and not really interacting that much

with the world made it feel really boring and time seemed to slow down going from each

objective.’ (P245)

‘Very few obstacles and dangers made the game a bit boring.’ (P142)

‘It felt very monotonous, with no combat or scares or music to really break up the flow of

the game and vary its pace.’ (P132)

Another possibility, as discussed in section 4.3.2, is that participants may have created

their own challenges in lieu of challenge provided by the game. This is again reflected by

participant comments provided in the survey sessions:

‘I set a challenge to myself to speedrun the game as quickly as possible (given my limited

skills).’ (P114)

‘I kinda made my own game out of juggling the canisters …’ (P113)

169


It may be that flow is still possible in the absence of designed challenge (or in experiences

in which the skills of the player outstrip the demands of the task, regardless of developer intention)

due to the potential for player-created challenge. This may signal limitations in the consideration

of flow as synonymous with optimal player experience; however, this finding also alludes to the

implicit difficulty of manipulating challenge‒skill balance in an experimental setting. A player

engaging with a boring or unchallenging video game at home has the option of simply

discontinuing play; as a laboratory session presents inherent expectations or pressures to play the

game, the player may instead contrive ways to make the experience more interesting—the

experience would thus not reflect typical play. In the context of industrial research, this limitation

could be avoided or minimised due to the accessibility of naturalised data from organic video

game players. The potential effects of this, as well as possible strategies for managing them, are

discussed in section 6.4. This may signal limitations in the consideration of flow as synonymous

with optimal play experiences; however, these conclusions are not meaningfully supported by data

evaluated within this program of research.

The lack of disparity in flow between the Balance and Boredom conditions despite the

removal of combat in the latter condition also strengthens the remaining arguments discussed in

section 4.3. The possibility remains that challenge‒skill balance may not be a necessary antecedent

to flow, or to all experiences of flow, as supported by Fong et al. (2014) and Rheinberg et al.

(2003). Another explanation may be that the game’s immersive quality is enough for a flow

experience, allowing for flow through absorption (Rheinberg et al., 2003). Finally, the S FSS and

FSS-2 scales may be inappropriate for assessing sub-optimal player experiences.

170


5.4.4 Player Experience of Needs Satisfaction Competence

While participants reported significantly less competence in the Overload condition than

both the Boredom and Balance conditions, no significant differences were found for competence

between the Boredom and Balance conditions. The lower experience of competence in the

Overload condition again aligns with the condition’s design intentions; as the challenge was

immediately overwhelming, and little progress could be made within the constraints of the

condition, it follows that participants would experience less competence within this play

experience.

The lack of significant differences between the Boredom and Balance conditions may be

explained by the PENS competence items used to assess the experience. As may have also

occurred with the FSS-2 and S FSS-2, some PENS competence items may be able to be rated

highly for both challenge‒skill balance and skill > challenge imbalance. For example:

[Item 2] I felt very capable and effective when playing.

The absence of sufficient challenge in the Boredom condition removes the opportunity for

failure or error. It may be that feelings of competence are thus not inhibited, allowing players to

experience mastery over the condition’s simple fetch task. It is unlikely that the participants may

have felt incapable, or ineffective, in the unimpeded retrieval of the gas canister objects.

Another item more directly addresses challenge‒skill balance, however, academic use of

the commercial PENS scale does not extend to publishing this item. The item enquires about the

player’s experience of challenge‒skill balance, but does not specify whether the absence of

balance is the result of challenge exceeding skill or skill exceeding challenge. It is possible that

participants responded to this item using either interpretation—that is, disagreeing with the item

could indicate both ability exceeding challenge, or challenge exceeding ability. As also discussed

171


in section 4.3.2, another possibility is that participants may have sought out their own challenge

for this condition—for example, attempting to collect all the fuel canisters before the condition

self-terminated.

Finally, experiences of competence in instances of sub-optimal or low challenge have

been identified in previous PENS literature. Described as ‘mastery in action’, this experience

occurs when players are granted the opportunity to ‘deliver a superlative performance without

having to work too hard’ (Rigby & Ryan, 2007). Importantly, Rigby and Ryan note that this is not

intended to represent the whole of the gameplay experience—rather, developers should provide

opportunities for mastery in action as well as optimally challenging play. Thus, it is possible that

the findings of the current study reflect participants’ experience of ‘mastery in action’ during the

Boredom condition, which resulted in high ratings of competence. However, this is arguably less

likely given that the whole gameplay experience (in the Boredom condition) was low challenge,

and ‘mastery in action’ is more likely to occur in situations where it is a ‘break’ from more

optimally challenging play.

All potential explanations point to implications for designers and player experience

researchers. Firstly, a high competence rating may not be indicative of challenge‒skill balance;

researchers should consider that players may feel that their skills outstrip the challenge in a game,

even when ratings suggest high levels of competence (unless the intention is to provide a ‘mastery

in action’ experience). Secondly, should the players be setting their own challenge within an

intentional skill > challenge imbalance condition, this highlights challenges associated with

designing a boring play experience while keeping the game artefact ‘game-like’ (that is, not

negating potential for invented challenge by removing interactive elements). Researchers and

172


designers may wish to evaluate whether players are creating their own challenges within a post-

game interview.

5.4.5 Player Experience of Needs Satisfaction Autonomy

Participants reported significantly greater autonomy in the Balance condition than in the

Boredom and Overload conditions, potentially positioning challenge‒skill balance as a precursor

to autonomy (or challenge‒skill imbalance as an inhibitor of autonomy). No differences were

found between the Overload and Boredom conditions. As a manipulation check, these results

position Balance as the optimal play experience in terms of autonomy.

In the Boredom condition, the linear task and absence of combat may have contributed to

reduced experience of autonomy: instead of being able to defeat enemy agents in interesting or

entertaining ways, participants were relegated to the repetitive task of canister collection.

Likewise, in the Overload condition, participants were almost immediately overwhelmed by a

large number of enemy agents, and thus constrained to the starting area of the map. Often,

participant movement was restricted either through incapacitation or enemy AI body-blocking. As

the enemy presence was continuous and overwhelming, it is also possible that the condition left

no room for strategy or target prioritisation, thus forcing the player to simply ‘spray and pray’

(shoot randomly at enemies).

It is notable that the Boredom condition did not elicit greater autonomy than the Overload

condition, as the Boredom condition provided participants with the ability to freely explore the

game world, choose their own routes and not remain near their AI teammates for safety. It is

possible that the presence of a clear challenge, even one that was overwhelming for the participant,

was able to balance this. Alternatively, the absence of an achievable goal in both the Boredom and

173


Overload conditions may have prompted participants to feel as though there were no meaningful

choices to be made.

As increases in autonomy are positively associated with increased intrinsic motivation

(Deci & Ryan, 2000), the presence of greater autonomy in the Balance condition may indicate

greater willingness to engage in the activity (Ryan et al., 2006). Through enabling a sense of

freedom and choice for players, the Balance condition was able to strengthen its position as the

optimal play experience of the three conditions.

5.4.6 Player Experience of Needs Satisfaction Presence

Finally, participants reported significantly greater presence in the Balance condition than

the Boredom and Overload conditions, with no significant differences revealed between the

Boredom and Overload conditions. As presence has been identified as a key factor in the

enjoyment of games (Lombard & Ditton, 1997), this again supports the Balance condition as the

most reflective of the optimal play experience among the three conditions.

These findings also position challenge‒skill balance as either a possible antecedent to the

experience of presence, or a strong predictor of it; conversely, this may also position challenge‒

skill imbalance as an inhibitor of presence, as supported by other research (Ravaja et al., 2004). It

is possible that the experience of presence was obstructed or interrupted within the Overload

condition due to unrealistic or notably skewed gameplay, thus taking players ‘out of’ the game

world. This may have occurred as a consequence of detachment from gameplay once participants

realised the insurmountable challenge of the game, thus diminishing mental or emotional

investment in the activity. The high rate of player deaths may have also contributed to the

interruption of presence, particularly as incapacitation and death removed participant control or

input for up to 15 seconds; participants were then returned to the starting area, where the process

174


began anew. In the event of repeated instances of player deaths, as occurred in the Overload

condition, participants may not have experienced uninterrupted play for long enough to evoke

experiences of presence.

Similarly, the complete absence of zombies or combat in the Boredom condition is at odds

with the story of the game world (as told by NPC commentary, world assets and prior knowledge

of the game), again potentially breaking the illusion of being transported to ‘within’ the game. It

is also possible that a lack of mental stimulation in the Boredom condition prompted

disengagement from the game, again inhibiting the potential for experiencing presence. These

findings are also supported by research undertaken by Ravaja et al. (2004), which identified ‘easy’

game modes as less likely to invoke feelings of presence. When designing for presence in the

game world, this points to the importance of obtaining challenge‒skill balance.

5.4.7 Confirmation of Condition Design Success

Generally, the challenge‒skill manipulation for the play conditions resulted in the

emergence of clear optimal and sub-optimal play experiences. Aligning with design intention, the

Balance condition was the most successful of the three conditions in evoking psychological

constructs that align with positive, or optimal, play experiences. The Balance condition was self-

reported by participants as the most successful in terms of attaining presence, autonomy and

interest/enjoyment. As for competence and flow, while no differences were revealed between the

Boredom and Balance conditions (potentially as a consequence of scale limitations), Balance still

emerged as more successful than Overload.

No results revealed the Boredom and Overload conditions as more successful than the

Balance condition in eliciting an optimal play experience, allowing for a direct comparison of

optimal (Balance) and sub-optimal (Boredom and Overload) conditions within the

175


psychophysiological results. Differences were found between the Boredom and Overload

conditions in some measures, with Boredom evoking greater experiences of competence and flow

than Overload, and Overload evoking greater interest/enjoyment than Boredom. These subjective

differences also assist in an exploration of the psychophysiological response to optimal and sub-

optimal conditions that differ between challenge‒skill balance and imbalance.

5.5 PSYCHOPHYSIOLOGICAL RESPONSE TO PLAY

The following sections explore RQs 1a and 1b by investigating the psychophysiological

response to the video game conditions. These also inform the overarching question of the utility

of psychophysiological measures in evaluating play, thus exploring the research aim by expanding

contemporary understanding of the value of psychophysiological measures in assessing the player

experience. While this section primarily examines the differences that emerged in the

psychophysiological response between the optimal and sub-optimal conditions, the potential for

psychophysiological measures as a predictor of the subjective experience—and potential

limitations in this approach—is also discussed.

5.5.1 Electrodermal Activity

Analysis revealed significantly lower levels of EDA in the Boredom condition than in the

Balance and Overload conditions, with no significant differences found between Balance and

Overload. The comparatively low levels of EDA in the Boredom condition may be unsurprising

in the context of the game condition—whereas both Balance and Overload featured combat,

violence, risk of failure and challenge, the monotonous nature of the Boredom condition entailed

only that participants traverse an enemy-free map to fulfil a repetitive fetch task. Dawson et al.

(2000) report that EDA, primarily a measure of arousal, is associated with stimulus novelty,

intensity, surprise, significance and emotional contentment; as most of these experiences are

176


notably absent in the Boredom condition, this plausibly explains the lowered EDA response for

the condition.

Despite the apparent intuitiveness of the result, the finding of decreased EDA in low-

stress, boring or unchallenging play experiences is not corroborated by research in the player

experience space. In research undertaken by Mandryk et al. (2006b) comparing beginner, easy,

medium and hard difficulties, no main effects of difficulty level were found on any of the

physiological measures, including EDA, employed within the study. Mandryk et al. suggest that

this may be a consequence of inconsistent participant responses to the difficulty settings, and—as

a result of methodology that featured consistent participant‒researcher interviews throughout—

response to the experimental situation, rather than to the experimental manipulations. It should be

noted that Mandryk et al. also report that only the beginner condition was perceived as

significantly less challenging than the remaining conditions; this may indicate a lack of clear

challenge distinction between conditions, or be a function of the relatively small sample size of

seven.

The findings of this research program also differ from those of Kneer et al. (2016), in

which no effect on physiological arousal (measured by EDA) as a consequence of difficulty was

found. It is possible that the difficulty manipulations used in this study are incomparable to those

used by Kneer et al.—for example, the Overload condition in this research was developed with

the intention of overwhelming the player and rendering the task impossible; Kneer et al. aimed for

higher difficulty, but possibly not to the same extent. This could also be true of the Boredom

condition in comparison to the low-difficulty condition employed by Kneer et al. Another

explanation for the disparity in results may be differences in experimental design—play times in

the study undertaken by Kneer et al. were 20 minutes in length, compared with the 10-minute play

177


sessions employed in this study. Therefore, it is possible that participants in Kneer et al.’s study

played long enough to habituate more readily to the high difficulty than in the Overload condition

used in this research.

Additionally, in Drachen et al.’s (2010) exploration of the physiological correlates of

player experiences in first-person shooter games, no significant correlation between EDA and

challenge—as measured by a self-report GEQ—was revealed. Drachen et al. propose that this was

a function of imprecise wording of the scale used, emphasising the difficulties of evaluating

challenge in player experience evaluation.

In Nacke and Lindley’s (2008) investigation of boredom, flow and immersion play

conditions, the boredom condition generated greater psychophysiological activity—including

EDA—than the immersion condition. Additionally, greater EDA was revealed in the flow

condition than in the boredom condition. These findings suggest that the relationship between

difficulty and EDA response may be moderated by additional variables; whereas the conditions in

the current research program varied only for challenge‒skill manipulation, Nacke and Lindley also

manipulated audiovisual and sensory experiences (through narrative, sensory effects and

environments). One conclusion could thus be that increased EDA indicates greater challenge, but

not necessarily a more optimal player experience.

Further clarification may be found in research undertaken by Ravaja et al. (2008), in which

increased EDA response was found when a player was killed or wounded, or when they killed or

wounded an enemy (Ravaja et al., 2008). In the Boredom condition, no enemies were present and

player death was impossible; in the Balance condition, some enemies were present, and player

death was possible; and finally, in the Overload condition, an excessive number of enemies were

present and player death was inevitable. Therefore, increased exposure to the death/wounding of

178


both enemy opponents and the player‒character may also be partially responsible for increased

EDA response. This may also potentially be true of the results reported by Nacke and Lindley

(2008).

The applicability of Ravaja et al.’s (2008) conclusions to this study may be complicated

by the absence of differences in EDA response between the Balance and Overload conditions. As

previously stated, participants experienced greater enemy numbers and higher incidences of player

death in the Overload condition than in the Balance condition; following these conclusions, an

expectation may be that EDA response would be significantly higher for the Overload condition

than the Balance condition. The absence of this difference, however, is possibly explained by the

potential for detachment or disengagement from the activity, or by habituation.

Participants in the Overload condition may have disengaged, or ‘given up’, during play

once the impossibility of completing the objective became apparent—as a result, the experience

may have been less affecting in terms of physiological arousal. Therefore, the tonic exploration of

the full 10-minute play session may present limitations in analysis, as participant arousal may have

only diminished once they reached the conclusion of impossibility—thus resulting in

electrodermal activation similar to those seen in the Balance condition. Anecdotal feedback

participants provided during the survey sessions corroborate this experience of eventual

detachment:

‘Having (what I assume) the difficulty turned up made it much more of a challenge, which

made me much more engaged with the task at hand, but as I kept dying, I started to become

less interested.’ (P245)

‘Once I realised the game was too hard I became more detatched [sic].’ (P243)

179


‘Held attention till realisation of imposible difficulty at which point and want to achive

victory and progression is lost [sic].’ (P223)

As discussed in section 2.8.6.1, habituation describes the process wherein

psychophysiological response is reduced after exposure to the continued presentation of, or

interaction with, the same stimulus (Stern, 2001, p. 55). This may be explained by the repetitive

experience of play in the Overload condition—players were almost overwhelmed by the enemy

zombies, died and had to respawn and repeat the scenario until the session terminated. Conversely,

in terms of study design, the semi-counterbalancing of the condition order may have resulted in

habituation to the conditions regardless of differences in challenge. As the Overload condition was

always placed last for fear of lasting effects on the Boredom and Balance condition, Overload may

be more susceptible to habituation—a possible explanation for the absence of differences in EDA

arousal between Balance and Overload.

A final explanation may simply be that challenge‒skill imbalance, wherein the challenge

outstrips the skill of the player, is equally arousing within video game contexts as challenge‒skill

balance. However, in the face of results gathered from other psychophysiological measures used

within this study, this conclusion seems unlikely.

5.5.2 Electrocardiography—Heart Rate

Analysis revealed significantly greater HR in the Boredom condition than the Overload

condition, with no significant differences revealed between the Boredom and Balance conditions.

The finding of greatest HR in the Boredom condition is seemingly at odds with the results gathered

from EDA, as both are measures of arousal—in particular, as with EDA, HR has been found to

increase during experiences of stress, mental activation and anxiety (Melillo, Bracale, & Pecchia,

2011; Allen et al., 1987; Szabo & Gauvin, 1992; Andreassi, 2007). However, as discussed in

180


section 2.8.6.4, physiological arousal does not extend along a unidimensional continuum from low

activation (deep sleep) to high activation (agitation, excitement or panic). As concluded by Lacey

(1967) and Stern et al. (2001), one form of arousal cannot always be used as a valid measure of

another form of arousal, nor be used to singly represent the psychological experience of an activity.

The principle of stimulus-response specificity states that specific stimulus contexts may

also bring about a specific pattern of physiological response, rather than a uniform increase or

decrease across a unidimensional continuum. Furthermore, directional fractionation (Lacey, 1967)

finds that physiological response does not decrease, or increase, uniformly across all measures. In

Stern et al.’s (2001) examples of both the missing wallet and the soldier on guard duty (discussed

in section 2.8.6.4), EDA increases even as HR decreases. It could be that, in the context of video

game play, play experiences may bring about a specific set of physiological response that features

directional fractionation—in that while EDA decreases in instances of low challenge or boredom,

HR increases.

The increased HR in the Boredom condition may be an anticipatory response. Although

HR is an autonomic activity, the frequencies, variations and pace of the contractions are responsive

to psychological stimuli of stressors, frustrations and fears (Andreassi, 2007, p. 438‒439); Melillo,

Bracale, & Pecchia, 2011). The increased HR may thus be a sympathetic response in preparation

for combat or threats—the presence of which is suggested by not only the context of the chosen

video game artefact (a well-known first-person shooter that typically features enemies

combatants), but also the environments within the video game conditions (a post-apocalyptic

world that contains multiple narrative and embedded suggestions of a zombie threat). This

explanation is further supported by comments provided anecdotally in participant feedback for the

Boredom condition:

181


‘There was still some excitement as I was half expecting zombies to turn up at any minute.’

(P140)

‘I think I learned quite quickly that the game was on it’s [sic] easiest mode and then began

to expect less shocks to almost not expecting any at all anymore towards the end.’ (P214)

‘I collected the can [sic], believing more enemies would spawn as you progressed

collecting the can …’ (P109)

Despite this, HR was lowered in the conditions that actually featured combat, threat and

risk of failure. As with EDA, this may also be a consequence of stimulus-response specificity—

in the context of the video game conditions, while HR increases in anticipation of a threat, actual

combat may result in HR decreases.

Another explanation may be that, due to the relatively decreased experience of presence

and interest/enjoyment (sections 5.4.6 and 5.4.2), participants were instead more conscious of their

environment and thus responding instead to the experimental manipulations and setting—as

experienced as a consequence of repeated interviews in research undertaken by Mandryk et al.

(2006b; discussed in section 2.9.1). Potential anxieties associated with the experimental setting

(unfamiliar laboratory environment, unfamiliar researcher and novel experiences) may thus have

contributed to the HR increase that occurred within the Boredom condition.

A speculative explanation for the pattern of results may point to the lowered HR in

Overload and Balance conditions as a consequence of different experiences (as discussed in

section 2.8.6.5, many-to-one and one-to-many domain relationships suggest that the same

physiological response be indicative of separate—and occasionally contrasting—psychological

experiences). Research undertaken by Nacke and Lindley (2008) revealed decreased physiological

arousal in an immersion condition than in a boredom condition; as immersion is conflated with

182


presence, the same may hold true in the context of HR results in this research program. If decreased

psychophysiological arousal is potentially associated with immersive game worlds, it is plausible

that the Balance condition evoked a lower HR as a consequence of its immersive environments.

Separately, participants playing the Overload condition may have also experienced a lower HR as

a consequence of habituation, or alternatively, disengagement in the face of overwhelming

challenge.

The discovery of a main effect of challenge‒skill balance on HR contrasts with research

undertaken by Mandryk et al. (2006b) and Kneer et al. (2016), in which no main effect of difficulty

on HR was discovered. Mandryk et al. suggest that an absence of results in their study may be the

consequence of the experimental procedure; however, in terms of the research undertaken by

Kneer et al., the disparity may lie in differences in play session lengths and condition design (as

discussed in section 2.9.2.1). Conversely, research undertaken by Drachen et al. (2010) revealed

correlations between increased HR and feelings of frustration and tension reported during play of

three first-person shooter games; this could support the potential for the physiological response to

be driven by anticipation in the Boredom condition, and disengagement in the Overload condition.

5.5.3 Electrocardiography Heart Rate Variability (High-frequency Peaks)

Significantly lower HF peaks were found in the Overload condition than in the Balance

and Boredom conditions, with no significant differences found between Balance and Boredom.

The lowered HF peaks in the Overload condition are congruent with psychophysiological

literature: HF reflects parasympathetic activity and has been found to decrease under conditions

of emotional strain and anxiety (Nickel & Nachreiner, 2003; Jönsson, 2007; Billman, 2013), with

decreases in general HRV associated with increases in stress states (Schubert et al., 2009). It is

183


thus reasonable to assume that the overwhelming difficulty of the Overload condition, coupled

with continued failure to progress, was capable of inducing emotional strain and stress.

The absence of differences in HF peaks between the Balance and Boredom conditions are

surprising due to the associations between decreases in the HF component of HRV and increased

mental workload and attentional focus (Cinaz et al., 2010; Hjortskov, N., 2004). An interpretation

for this may be that the Boredom condition, despite the monotony of its experience, evoked similar

levels of attentional focus or mental workload as the Balance condition due to anticipatory

response, as discussed in section 5.5.1. Additionally, as Balance and Boredom did not differ in

terms of self-reported flow, mental workload may have been equally diminished (assuming these

self-reported results represent actual experience): several of the flow dimensions describe a

merging of action‒awareness, loss of self-consciousness and a sense of control over the activity.

However, this argument is weakened somewhat by the presence of an additional dimension

concerning intense and focused concentration

These results fail to align with those of previous research undertaken by Keller et al.

(2011). In Keller et al.’s investigation of the psychophysiological response to flow as manipulated

by challenge‒skill balance, decreases in HRV indicating enhanced mental workload were reported

in a challenge‒skill balance (‘fit’) condition. Accordingly, Keller et al. posit that physiological

elements that reflect tension and mental load may also indicate flow in the player experience.

However, it is critical to note that two substantial distinctions between the study and interpretation

approach may account for the disparity in results. A careful reading of the study analysis shows

that the difference between the fit and overload conditions was not significant, with the p-value (p

< .10) interpreted by Keller et al. as ‘trend-level significance’; in this program of research,

however, differences not achieving significance are not interpreted. An additional key difference

184


is found in the choice of the game artefact for study: Keller et al. employed a game based on the

television show Who Wants to be a Millionaire, featuring quiz-format questions participants were

required to answer from a multiple-choice selection within a limited time period. As discussed in

section 2.9.6, the generalisability of this artefact within the space of player experience is

potentially limited. Furthermore, the quiz-like nature of the chosen game artefact is likely to differ

in terms of mental workload from that experienced in the action first-person shooter selected for

this research program; arguably, in the video game condition selected by Keller et al., explicitly

requiring increased mental workload was the game’s primary function.

5.5.4 Electromyography Orbicularis Oculi

Analysis revealed that the Overload condition featured significantly higher levels of EMG

OO activity than either the Balance or Boredom conditions. Additionally, the Balance condition

showed significantly higher levels of EMG OO activity than the Boredom condition. As EMG OO

is primarily employed as a measure of positively valenced emotion through activation of the OO

muscle that occurs when smiling (Andreassi, 2007, pp. 300–303), these results present interesting

implications for the role of EMG OO in evaluating player experience.

The most intuitive result could be the relative lack of EMG OO activity in the Boredom

condition when compared with the Balance and Overload conditions. Again, the monotony of the

Boredom condition is potentially a root cause for the reduced activity; as the only task available

to the players in this session was the repetitive retrieval of gas canisters, there was a notable

absence of events capable of provoking an emotional experience positive enough to elicit a smiling

response. This finding is supported by research undertaken by Nacke and Lindley (2008), in which

the flow condition—featuring challenge‒skill balance—prompted greater EMG OO activity than

the boredom condition. Notably, in an exploration of physiological associations with flow

185


experiences, Kivikangas (2006) was unable to find associations between flow and EMG OO; while

the current research program is unable to be directly compared with specific investigations of flow

experiences, the EMG OO results reported herein may further expand on the current understanding

of physiological response to optimal play conditions.

Of particular interest is the greater activity of EMG OO in the Overload condition than

the Boredom condition. It is feasible that, owing partially to the immersive nature of the Balance

condition and the interrupted nature of the Overload condition, participants were more

expressively responsive in the Overload condition. To extrapolate on this, player deaths within the

game world allowed for up to 30-second ‘breaks’ from play that were not also present in the

Boredom and Balance conditions. During these breaks, participants could emote in response to

the level’s overwhelming difficulty or the participant player‒character’s death in the game; these

expressions could have been the result of exasperated, nervous or amused laughter, either as a

consequence of stress relief or pleasure, the presence of which is anecdotally supported by

feedback provided in the survey session for the Overload condition:

‘[I] couldn't help but laugh before and during the game at the difficulty.’ (P130)

In addition to the above participant comment, this interpretation is further supported by

researcher observation of the participants during the study.

Another possible explanation for the Overload results may stem from the tonic analysis

limitations. As the phasic assessment of EMG is not reliably feasible within commercial contexts

due to the associated time costs, and tonic analysis of EMG has been previously employed in

player experience literature (Kivikangas, 2006; Nacke & Lindley, 2008), EMG OO was assessed

tonically within this program of research. However, it should be noted that EMG OO is typically

assessed within psychophysiological spaces through the evaluation of phasic reactions to a single

186


stimuli (Ekman, Davidson, & Friesen, 1990), and is an intended avenue of exploration in future

research.

A disadvantage of tonic analysis is the potential for conflating startle response with

psychophysiological results (see section 2.8.6.3). The magnitude of startle response—defensive

eye-blink reflexes to unexpected or threatening stimuli, capable of being recorded as EMG OO

activity—is inversely related to the valence of the stimuli (Lang, 1995). In the experience of the

Overload condition, it is plausible that the high presence of threat, tendency for enemy combatants

to attack from behind and sudden deaths may have resulted in higher incidences of startle response

than either the Boredom or Balance conditions. Therefore, despite the interpretation of EMG OO

as a measure of positive valence, it is possible that the OO activity is also conflated with

experiences of negative valence through startle response. In their exploration of the

psychophysiological experience of flow, Kivikangas (2006) identifies this as a possible limitation

for interpreting their results.

Finally, player experience of the Balance condition is associated with higher levels of

EMG OO than Boredom and lower levels of EMG OO than Overload. If the EMG OO activity of

Overload is not conflated with startle response, this suggests that researchers and developers

evaluating player experience should be cautious of interpreting high OO activity as necessarily

indicative of positive or optimal experiences. These implications for using EMG OO in the

assessment of the player experience are further discussed in section 6.5.

5.5.5 Electromyography Corrugator Supercilii

No significant results were obtained in the analysis of EMG CS. This is likely a

consequence of the notably reduced n available for evaluating this measure, as a probable

187


consequence of signal loss or unresolvable noise levels (see section 5.1.16). Recommendations

for future methodologies employing this measure are suggested in section 6.4.3.

5.5.6 Electroencephalography

A similar pattern of results was found for all measures of EEG that achieved significance:

in the evaluation of the absolute power of AF4 Beta, O2 Alpha, O2 Beta and O2 Theta, activity in

the Balance condition was found to be significantly lower than activity in the Boredom condition.

In all instances, no differences were revealed between Balance and Overload. This section

discusses a potential cause for this similar pattern of results, and offers interpretations for the

results as they occurred.

5.5.6.1 AF4 and O2 Beta

The most intuitively interpretable findings are the pattern of results revealed for the AF4

Beta and O2 Beta frequency bands. As beta activity occurs during states of alertness, it is most

common when an individual is engaged in mental activity (Andreassi, 2007, p. 69), and is

associated with cognitive task demands, information processing and problem-solving (Fernandez

et al., 1995; Cole & Ray, 1985; Nacke, 2010).

In terms of the Boredom condition, beta activity reduction may occur as a consequence of

reduced mental simulation through monotony and task repetition; the relatively simple fetch

task—moving towards a highlighted object on the map, collecting it and returning it to the same

designated location—does not represent a cognitively demanding task, nor allow room for

problem-solving beyond determination of the shortest routes to the highlighted objects. Reduced

AF4 and O2 Beta in the Boredom condition can thus be explained by a dearth of cognitive

stimulation.

188


The Balance condition is intuitively facilitative of increased beta activity as a consequence

of challenge‒skill balance. Unlike the Boredom condition, the Balance condition necessitates

problem-solving and information processing in order to successfully complete the games

objectives. The opportunities for problem-solving present in a multitude of ways: special enemies

(see section 3.3.4) require unique strategies to defeat; some enemies require immediate

prioritisation over others due to proximity to the player‒character; the participant must decide

whether they need to drop a canister to shoot the enemy, or attempt evasive action; managing

health and ammunition resources is necessary; panic events—in which multiple enemies appear

at once—require the player to play defensively; and so on. As the cognitive demands increase,

mental activity, problem-solving and information processing also increase to meet the demands.

The increased beta activity in the Overload condition, with no differences revealed

between Overload and Boredom, could suggest the same conclusions introduced in section 5.5.2:

players may have disengaged at a certain point within play, but not immediately. Increased beta

activity may thus occur in the Overload condition for the same reasons as the Balance condition

(problem-solving, cognitive demand and information processing). However, as physiological

activation does not take place on a single unidimensional continuum, beta in the Overload

condition failed to outstrip the Balance condition as a consequence of impaired engagement at

some point during play.

Within the context of electroencephalographic studies of the player experience, beta

activity has also been associated with feelings of spatial presence (Nacke, 2010; Johnson et al.,

2015). As the Balance condition was also found to score highest on PENS presence, this may

represent an additional influence on the increased beta activity in Balance—further supporting

EEG Beta activity as indicative of video games presence. Despite this, a predictive or correlational

189


relationship was not shown between PENS presence and EEG Beta, although this may be a

consequence of variances; see section 5.3.6 for discussion on this.

5.5.6.2 O2 Theta

As revealed in section 2.8.7.4, the psychological stimuli responsible for theta activity is

less clear-cut than beta. Theta has been found to occur in both states of drowsiness, sleep, problem-

solving and attentional focus (Stern et al., 2001, p.81), as well as daydreaming, creativity,

emotional processes and working memory tasks (Mitchell et al., 2008). Within player experience

literature, relationships have also been established between theta activity and challenge or

engagement (Salminen & Ravaja, 2007).

The obvious connection, and potentially the most reasonable explanation for these results,

is that of increased theta activity with challenge and engagement. The play conditions were

directly manipulated for challenge, such that the Balance condition emerged as the optimal play

experience through challenge‒skill balance and Overload as sub-optimal through challenge > skill

imbalance. The relatively decreased theta activity in the Boredom condition may again be

explained by disengagement as a consequence of monotony, with absence of challenge in the

Boredom condition occurring due to the removal of combat. If these assumptions reflect the

participant experiences, the increased theta activity in the Balance and Overload conditions aligns

with extant player experience literature.

Other explanations for the increased theta activity in Balance and Overload emerge

through theta associations with creativity, emotional processing, attentional focus and working

memory tasks. A state of creativity and attentional focus may occur as a consequence of strategy

and problem-solving; furthermore, as the Balance condition allows for more ‘highs and lows’—

rather than being consistently monotonous or consistently overwhelming to the point of

190


disengagement—emotional processing may also be at its peak in the Balance condition. Finally,

working memory may occur as a consequence of remembering how to best approach the special

zombie types, as each zombie has its own unique attack and strategy; these solutions are

unserviceable in the Boredom condition due to the absence of enemy units.

Overall, these conclusions suggest that theta activity in video game play is unlikely to

indicate the restful states also associated with theta activity. This may be because, regardless of

the level of game challenge, simply playing a video game and acting within game environments

could inhibit theta activity associated with drowsiness.

5.5.6.3 O2 Alpha

Alpha activity (EEG in the 8-12 Hz range) is observed in individuals when they are awake

but relaxed, a state of ‘relaxed wakefulness’ (Davidson et al., 2000, p. 31), and is particularly

associated with the closing of the eyes. Alpha activity is associated with a relative lack of cognitive

processing (Stern et al., 2001, p. 80), with the presence of alpha indicative of ‘idling’ (Schier,

1999). Increases in alpha activity are typically interpreted as indicative of less attentional activity,

with decreases indicative of more attentional capacity. This is further emphasised by the Berger

effect, wherein the amplitude of the alpha band decreases notably when the subject opens their

eyes (Bazanova & Vernon, 2014). Complex cognitive tasks, or the introduction of a stressor,

interrupt increased alpha states in a process known as ‘alpha blocking’ (Stern et al., 2001, p. 80).

Within player experience research, alpha has been found in at least one paper to decrease with

positive mood (Russoniello et al., 2009).

The increased alpha observed in the Overload condition is potentially attributable to

disengagement or habituation due to the overwhelming task demands. This disengagement may

have resulted in a relaxed state, or less attentional activity, increasing the presence of theta activity.

191


This could also explain why increased alpha has emerged alongside increased beta and theta

activity, as these mental states may have occurred at different stages of play within the Overload

condition. Additionally, if the relationship observed by Russoniello et al. (2009) hold true,

increases in negative mood as a consequence of repeated failure—thus increasing frustration or

anxiety—may also be responsible for increases in alpha activity in the Overload condition.

That alpha activity was significantly greater in the Balance condition than in the Boredom

condition is surprising. It was expected that increased alpha, indicative of increased relaxation,

might be associated with the Boredom condition. One explanation may be that, as the Balance

condition featured combat and risk of failure, more emotional ‘highs and lows’ occurred than in

the Boredom condition; as such, O2 Alpha may have increased with negative mood. Jennet et al.

(2008) hypothesises that faster paced games also lead to greater negative affect than do slower

paced games, further supporting the notion that the Balance and Overload condition evoked

greater negative emotional experiences than did the Boredom condition. Despite this, increased

alpha activity is typically associated with less attentional activity (Schier, 2000); it is possible that,

in the specific context of gameplay, alpha activity may be more indicative of negative emotional

experiences than of attentional inactivity. As such, future research may expect to find increased

alpha activity in more emotionally or cognitively challenging play experiences.

Another explanation may be that increased alpha in the Balance condition represents a

state of ‘peak flow’, or relaxed—but pleasurable—engagement. This doesn’t aid in understanding

of the increased alpha results in the Overload condition, but (as established in section 2.8.6.5) the

same physiological response is not necessarily indicative of the same psychological processes.

192


Despite these possible explanations, these findings remain perplexing. The large effect

sizes associated with the EEG results suggest that this is an effect to pursue in future research—

especially if the counter-intuitive results are indicative of optimal states.

5.5.6.4 Shared Pattern of Results

One potential explanation for the shared pattern of results among all frequency bands and

both sites may be the limitations of the equipment used for recording. The EEG instrument used

for this study, a 14-sensor dry array EMOTIV Epoc EEG headset, was primarily designed for

‘practical’ research in commercial and research spaces. In Duvinage et al.’s (2013) comparative

analysis of the EMOTIV Epoc EEG headset with a medical grade counterpart, the Epoc headset

was found to be limited in its performance as a consequence of inconsistent positioning, greater

potential for movement artefacts and a greater chance for misinterpretation of cognitive activity.

Despite this, Duvinage et al. state that the EMOTIV Epoc headset is ‘not bad at all for such a low-

cost system’, with responsiveness to participant cognition far above the chance level of 25%.

Furthermore, Duvinage et al. recommend the headset for use in non-critical contexts, such as

games. It should be highlighted here that this analysis employed the EMOTIV Epoc as an

instrument for interaction between the brain and computer, and not for frequency analysis as

undertaken within this program of research.

As a potential consequence of these issues, the similar pattern of results may point to

issues in evaluating the absolute power of individual frequency bands with the chosen analysis

method or equipment—results could thus simply be interpreted as general cognitive activity. This

may indicate that Boredom was less cognitively demanding overall, as a consequence of the

monotony, repetition and ease of the condition. Furthermore, the absence of differences between

193


the Balance and Overload conditions could indicate similar levels of cognitive activity or

disengagement at some point during play of the Overload condition.

5.6 EFFECT SIZES

A low- to medium-effect size was revealed for many of the psychophysiological

measures; in particular, EDA, HF and HR revealed effect sizes between .072 and .102 (see Table

9 for all effect sizes). However, more relevant for this research program is the comparison of effect

sizes, which allows for the discernment of useful psychophysiological measures to explore optimal

and sub-optimal conditions (particularly those that vary for challenge‒skill balance).

Table 9. Effect sizes for psychophysiological measures

EDA ηp2 = .081

HR ηp2 = .071

HF Peak ηp2 = .102

EMG OO ηp2= .279

EEG AF4 Beta ηp2 = .137

EEG O2 Alpha ηp2 = .336

EEG O2 Beta ηp2 = .389

EEG O2 Theta ηp2 = .366

Overall, EMG OO and the EEG O2 site emerged as having the largest effect size for

differences between play conditions. This situates both measures as the most effective, among the

psychophysiological measures used within this research program, for detecting differences

between optimal and sub-optimal play experiences. Interestingly, EMG OO and the O2 site offer

effect sizes comparable to, and greater than, effect sizes reported for subjective evaluation (see

Table 10).

194


Table 10. Effect sizes for all subjective measures

Flow ηp2 = .282

Interest/Enjoyment ηp2 = .277

Competence ηp2 = .492

Autonomy ηp2 = .237

Presence ηp2 = .251

With the exception of competence, EEG O2 appears more effective at detecting

differences between the play conditions than subjective analysis. Additionally, with the exception

of competence and flow, EMG OO also emerges as a more robust tool for determining experience

differences between conditions.

The magnitude of the differences revealed by EEG OO is potentially a consequence of

O2’s proximity to the occipital lobe. Many regions of the occipital lobe are specialised for tasks

related to visuospatial processing, motion perception and receiving visual information. The

presence of enemy combatants in the Overload and Balance conditions may thus explain the

significantly increased activity in the O2 region for these conditions.

Furthermore, the comparatively strong effect size found for EMG OO provides additional

support for a relationship between increased challenge and increased EMG OO activity; this

finding depicts EMG OO as a particularly robust approach for evaluating differences between low

challenge and high-challenge player experiences.

The low effect sizes revealed for the remaining psychophysiological measures do not

lessen the importance of the results. As psychophysiological assessment is constrained to operate

within parameters normal for physiology, small effect sizes may not be indicative of weak results.

However, these findings do potentially indicate that EDA, HF Peaks, HR and the AF4 site are less

capable of discerning granular differences between conditions than EEG O2 and EMG OO in the

195


specific context of challenge manipulation; it should be noted, however, that these results may

differ in the manipulation of different stimuli (for example, violence, sound, or graphical

representations).

6 SUMMARY AND CONCLUSIONS

This program of research has explored the psychophysiology of the player experience, as

well as the applicability of the psychophysiological method to player experience evaluation,

through a large-scale approach incorporating a large sample size and the contemporaneous

employment of multiple physiological measures. Both subjective and psychophysiological

measures were assessed separately. The pattern of results for the subjective measures confirmed

the successful creation of optimal and sub-optimal experiences, and differences emerged between

play conditions for EDA, EMG OO, ECG (HR), ECG (HF peaks), AF4 Beta, O2 Alpha, O2 Beta

and O2 Theta.

6.1 SELF-REPORT SUMMARY

In terms of subjective experience, the play conditions were found to largely follow the

expected pattern of results, with Balance emerging as the most positively received play condition.

While this largely confirmed the success of the play conditions in evoking optimal and sub-optimal

play experiences, some surprising results became apparent that suggest both the difficulties of

player experience analysis and unexpected player preferences. These results expand on existing

knowledge of the player experience and player experience evaluation methods. See Table 11 for

an overview of the results for the self-report measures.

In all self-reported measures, Balance emerged as the most positively received condition.

For the flow and competence measures, the Boredom condition was received as positively as the

196


Balance condition, but Balance was at no point outstripped by either the Boredom or Overload

conditions. Overall, this establishes the Balance condition as the optimal play experience and

confirms the success of the condition design. This finding highlights the importance of challenge‒

skill balance in ensuring an optimal or positive play experience, as supported by a meta-analysis

of literature featuring challenge‒skill manipulations undertaken by Fong et al. (2014). However,

it is critical to note that this conclusion may be influenced by the expertise of the sample, with

participants self-rating as 5.96 out of 7 for ‘general experience with video games’; in previous

research, the role of challenge and challenge‒skill balance has been theorised as less important for

novice players (Lomas et al., 2013; Alexander et al., 2013).

As discussed in sections 5.4.3 and 5.4.4, the high ratings for flow and competence in the

Boredom condition may indicate complexity in the evaluation of easy, or skill > challenge

imbalanced, player experiences. Two theories were posited for this: either the scales for assessing

flow and competence did not allow for the distinction between challenge‒skill balance and skill >

challenge balance experiences, or the potential for player-created ‘fun’ or challenge must be

considered in future evaluation of player experiences.

A final interesting note is the performance of the Overload condition in terms of

interest/enjoyment. The Overload condition was significantly more positively received than the

Boredom condition, potentially indicating player preference for challenge > skill imbalance over

skill > challenge imbalance. As with the reception of challenge‒skill balance, however, this may

be influenced by the expertise of the sample collected for this program of research.

197


Table 11. Overview of significant results for subjective measures

S FSS - Flow (ηp2 = .282)


Boredom ‒ BO > OL**

Balance ‒ BA > OL**

Overload BO > OL** BA > OL** -

IMI: Interest/Enjoyment (ηp2 = .277)


Boredom ‒ BA > BO** OL > BO**

Balance BA > BO** ‒ BA > OL**

Overload OL > BO** BA > OL** -

PENS: Competence (ηp2 = .492)


Boredom ‒

Balance BA > OL** ‒ BA > OL**

Overload BO > OL** BA > OL** -

PENS: Presence (ηp2 = .251)


Boredom ‒ BA > BO**


Overload BA > OL** -

PENS: Autonomy (ηp2 = .237)


Boredom ‒ BA > BO**


Overload BA > OL** -

* p < .05, ** p < .01

Boredom = BO Balance = BA Overload = OL

198


6.2 PSYCHOPHYSIOLOGY SUMMARY

Analysis of the physiological response to the condition revealed results both congruent

and incongruent with existing literature within the player experience domain (see Table 12 for an

overview of these results). Potential explanations for this complicated pattern of results, rooted in

both psychophysiological and player experience research, are offered in more comprehensive

detail in section 5.5. Overall, these results contribute to current understanding of the

psychophysiological experience of play and offer insight into psychophysiological methodologies

in player experience evaluation contexts.

Measures of physiological arousal revealed interesting, novel and occasionally

incongruent results. The Overload condition featured significantly increased EDA and lower HF

peaks than the Boredom and Balance condition; this contrasts with previous psychophysiological

literature in the player experience space, in which no main effect of difficulty on either of these

measures was found (Mandryk et al., 2006b; Keller et al., 2011; Kneer et al., 2016). This research

program reveals that EDA and HF response to game difficulty can emerge, corresponding with

domain-level psychophysiological literature associating these physiological responses with stress,

tension and anxiety (Dawson et al., 2000; Nickel & Nachreiner, 2003; Jönsson, 2007).

Despite this, the Boredom condition featured a significantly higher HR than the Overload

condition—as increased HR is also associated with high arousal, stress and emotional intensity

(Andreassi, 2007; Melillo et al., 2011), this result seemingly contradicts the pattern of EDA and

HF results revealed in the Overload condition. Furthermore, the incongruity between lower peak

HF in the Overload condition and increased HR in the Boredom condition is surprising. A timebin

analysis of the first and last five-minute windows of both the Boredom and Overload conditions

would yield a richer understanding of this incongruity; this is further discussed in section 6.5. An

199


additional explanation, as offered in section 5.5.1, may be that habituation or disengagement in

the Overload condition could specifically explain relatively higher HR in the Boredom condition.

The disparity might also be explained more generally by stimulus-response specificity, in that

physiological response does not increase uniformly across all measures.

The EMG OO findings somewhat corroborate player experience research undertaken by

Nacke and Lindley (2008), in that OO was lowest in the boredom condition employed in their

research. Generally, the findings reported in this research program may imply that challenging

player experiences are more capable of inducing OO activity than unchallenging or easy player

experiences. This develops current understanding of tonic assessment of EMG in player

experience literature.

The findings for the EEG AF4 and O2 Beta frequency bands may suggest decreased

mental stimulation in the Boredom condition as a consequence of monotonous or repetitive

experiences; furthermore, mental stimulation, information processing and problem-solving are

associated with the increased challenge experiences of the Balance and Overload conditions. As

presence has been associated with beta activity in previous player experience research (Nacke,

2010), this may also explain the increased beta activity in Balance (rated highest on PENS

presence). The increased O2 Theta in Balance and Overload supports previous research

undertaken by Salminen and Ravaja (2007) in which theta was associated with game challenge;

alternatively, theta associations with creativity, emotional processing, attentional focus and

working memory tasks may also explain these results. The conclusions also support the notion

that theta activity in video game play is unlikely to indicate the restful states also associated with

theta activity. Finally, the findings for increased O2 Alpha in Balance and Overload may be

attributable to the improved potential for emotional ‘highs and lows’, otherwise inhibited by the

200


monotony and task repetition of the Boredom condition. Disengagement or habituation throughout

the Overload condition may also explain the increased alpha activity.

Despite these findings, one explanation for the EEG findings—as a consequence of the

shared pattern of results among all frequency bands and sites—is that more cognitive activity

occurred during the Balance and Overload conditions. However, interpreting the individual

frequency bands prompts potential useful insights into the psychophysiological response to play.

201


Table 12. Overview of significant results for psychophysiological measures

EDA (ηp2 = .080)


Boredom ‒ OL > BO**

Balance ‒

Overload OL > BO** ‒

HR (ηp2 = .072)



Balance ‒

Overload BO > OL** ‒

HF Peak (ηp2 = .102)



Balance ‒ BA > OL*

Overload BO > OL** BA > OL* ‒

EMG OO (ηp2= .279)



Balance BA > BO** ‒ OL > BA**

Overload OL > BO** OL > BA** ‒

EEG AF4 Beta (ηp2 = .136)


Boredom ‒ BA > BO** OL > BO*

Balance BA > BO** ‒

Overload OL > BO* ‒

EEG O2 Alpha (ηp2 = .335)





EEG O2 Beta (ηp2 = .389)





EEG O2 Theta (ηp2 = .368)





* p < .05, ** p < .01

Boredom = BO Balance = BA Overload = OL

202


6.3 EXPLORATION OF RESEARCH QUESTIONS AND AIM

This program of research sought to identify the effectiveness of psychophysiological

measurement in player experience evaluation. To explore this, a large-scale psychophysiological

study was undertaken to both investigate the utility of various physiological measures and assess

the psychophysiological response to optimal and sub-optimal player experiences. This section

explores the research questions and aim in the context of the results discussed in this chapter.

Research Questions

1. How effectively can psychophysiological measures be used to evaluate the player

experience?




measures, most reliably predict specific components of the player experience

as assessed by subjective measures?

Initial analysis confirmed the successful creation of an optimal and sub-optimal play condition,

allowing for the exploration of RQ1a, although some results indicate challenges associated with

evaluating non-optimal player experiences. In the psychophysiological assessment of the

differences in psychophysiological response between optimal and sub-optimal play experiences,

several differences were revealed (for an overview, please see Table 12): the Overload (sub-

optimal in that challenge > skill) evoked greater EDA than the Boredom condition (sub-optimal

in that skill > challenge), greater EMG OO activity than both the Boredom and Balance conditions

and greater AF4 Beta, O2 Alpha, O2 Beta and O2 Theta than the Boredom condition. The Balance

condition (challenge‒skill balance; optimal condition) evoked higher HF peaks than the Overload

203


condition, and greater EEG AF4 Beta, O2 Alpha, O2 Beta and O2 Theta than both the Overload

and Boredom conditions. Finally, the Boredom condition evoked a higher HR than the Overload

condition and higher HF peaks than the Overload condition. Interpretations of these results, and

what they may reflect about the player experience, are discussed in section 5.5.

RQ1b required the assessment of predictive relationships, employing the

psychophysiological measures as the predictors and specific components of the player experience

(the subjective measures) as the outcome. As multiple regressions analysis revealed no significant

regressions equations, and correlations testing very few significant correlations, no

psychophysiological measures were found to reliably predict specific components of the player

experience. The reasons for this are offered in section 5.3.6, but broadly, this may establish

psychophysiological measures—particularly when restricted to a feasibly obtainable sample

size—as limited in the granular prediction of specific psychological concepts such as flow,

interest/enjoyment, presence, competence and autonomy within the context of challenge-skill

balance manipulation.

Aim: To further clarify existing contributions to literature by expanding understanding

of the psychophysiological experience of play, and the value of psychophysiological measures as

a means of assessing the player experience.

Ultimately, this program of research has addressed the aim of expanding current

understanding of the value of psychophysiological measures in assessing the player experience.

The psychophysiological responses to optimal and sub-optimal player experiences revealed within

this research program has expanded upon extant psychophysiological player experience literature;

in particular, it has revealed new findings regarding the physiological effect of challenge.

Psychophysiological assessment has been supported by this research as a useful and insightful

204


approach for player experience evaluation, and particular recommendation is made for its

employment alongside subjective analysis. The findings of this research program also contribute

methodological recommendations for using psychophysiological analysis within player

experience evaluation contexts.

6.4 APPLICABILITY OF PSYCHOPHYSIOLOGICAL ASSESSMENT

One aim of this research program was to assess the utility of psychophysiological

measures and methodologies in their application to player experience evaluation in both academic

and commercial contexts. As this research represents one of the first large-scale

psychophysiological evaluations of the player experience, this section largely discusses this

applicability in the context of the data collection, treatment and analysis of Study 2. This section

is additionally informed by the research program limitations, and existing psychophysiological

research in player experience. Recommendations are made for psychophysiological evaluation in

future research.

6.4.1 Time Costs

A notable disadvantage of psychophysiological assessment is the temporal cost associated

with experimental set-up, data treatment and data analysis. This program of research already

sought to minimise time requirements by employing tonic, as opposed to phasic, analysis of every

psychophysiological measure (including contexts where phasic assessment is typically used; see

section 5.5.4). Despite this, time costs associated with even tonic evaluation may prove prohibitive

within academic and commercial player experience research contexts. However, there are

solutions available to further minimise time requirements.

205


The contemporaneous use of EDA, EMG OO, EMG CS, EEG and ECG markedly affected

the experiment runtime, adding a total of 40 minutes (of mandatory participant presence) to the

experiment set-up. As participants cannot freely move during set-up or data collection, the entire

experimental process required that participants remain seated for a total of two hours—an

experience that some described as ‘uncomfortable’. The maintenance of EL254 EMG and ECG

electrodes further contributed to the time costs of each experiment session, as the electrodes

required scrubbing in tepid water and a five-minute exposure to disinfectant to both sterilise them

and prevent the potential for remaining electrode gel to dry within the electrode cup cavity.

Considerable time costs are associated with the treatment and analysis of

psychophysiological measures. In all instances within this program of research, all physiological

data was individually visually scanned for the presence of movement artefacts, signal interruption

and data loss. In the event of an artefact, it was necessary to match the timestamp available in the

numerical printouts with the occurrence of the artefact in the physiological trace; the compromised

data corresponding with the timestamp was then removed from analysis. Including the steps taken

to analyse and extract the data from the physiological recording software, each physiological

measure required between 10 and 20 minutes of treatment per participant. This process was further

extended by the rendering time required for frequency analysis of EEG, in which each site required

seven minutes of computer processing time to output the data for all frequency bands per

participant. This totalled 21 hours of computer processing time for frequency analysis on two sites

alone; in this program of research, this procedure was carried out on four high-end PCs to reduce

this time cost. Finally, all data required conversion to a readable format and layout for analysis

within SPSS; as each measure represented hundreds of cells of data across three conditions, this

process also represented a notable temporal cost. The estimated total time invested in non-

206


statistical analysis, treatment and set-up of all physiological measures in the current program of

research was approximately 300 hours. The breakdown of this is available in Table 13.

Table 13. Overview of temporal costs for psychophysiological measures

In the interest of minimising time cost for future research in both academic and commercial

contexts, some measures were found to be less time-consuming in set-up and analysis than others.

EDA ECG EMG

OO

EMG

CS

EEG

AF4

EEG

O2

Total

AcqKnowledge

Analysis

5 mins 10 mins 5 mins 5 mins 10 mins 10 mins ‒

AcqKnowledge

Render Time

‒ ‒ ‒ ‒ 7 mins 7 mins ‒

Kubios

Analysis

‒ 5 mins ‒ ‒ ‒ ‒ ‒

Artefact

Checking

10 mins 5 mins 15 mins 15 mins 20 mins 20 mins ‒

Analysis Set-

up

5 mins ‒ 5 mins 5 mins 15 mins 15 mins ‒

Total per n 20 mins 20 mins 25 mins 25 mins 52 mins 52 mins 194

mins

Sample Total 30 hrs 30 hrs 37.5 hrs 37.5 hrs 78 hrs 78 hrs 291 hrs

207


Furthermore, resources are available—although they were not employed in this research program

due to financial or methodological concerns—that also reduce time costs.

In terms of epoch, HR and HF analysis, EDA and ECG proved the most time-efficient

measures; in terms of EDA and ECG, movement artefacts are more instantly recognisable than

artefacts that may occur in EMG or EEG data, in which some artefacts may be obfuscated by

natural oscillations of the trace (EEG) or emoting (EMG). ECG movement artefact analysis is

further aided by the use of Kubios software, which features a rigorous automatic artefact correction

algorithm and respiration frequency analysis that ensures the HF component remains within the

HF band limits (Tarvainen et al., 2013)—thus limiting the need for manual removal of minor

artefacts.

While disposable electrodes were not used for EMG and ECG analysis within this research

program, the use of disposable electrodes would minimise the time costs associated with electrode

maintenance and preparation. Pre-filled disposable electrodes would also minimise the risk of air

bubbles in freshly applied electrode gel compromising the physiological data.

Finally, both software and outsourcing is available for detecting movement artefacts.

While these approaches were not employed in this research program due to financial limitations,

automatic movement artefact detection software or outsourcing to laboratories that offer this

service would also minimise the time cost of psychophysiological data in contexts where it is

accessible.

6.4.2 Viable Psychophysiological Approach

This program of research investigated the utility of multiple psychophysiological

measures in the assessment of the player experience. In an effort to address the primary research

question and research aim, each measure will be critically assessed in terms of interpretability,

208


temporal costs, effect size, and accessibility within the context of their use in this research. It is

intended that this assessment will prove helpful in informing experimental design for researchers

and developers considering a psychophysiological approach in future player experience

evaluations. For an overview of this discussion, please refer to Table 14.

Table 14. Overview of utility of psychophysiological measurements

Overall, EDA proved the most efficient measure in terms of interpretability and temporal

cost. The EDA measure also featured the lowest rate of data loss within this program of research,

with only four of 90 samples deemed unsuitable for analysis; this situates palmar EDA as

particularly robust to data contaminants or dislodgement. The relative ease, deployability, and

Interpretability Time Cost (per n)* Effect Size(s)

EDA Relatively

interpretable; some

findings contradict

existing research

25 mins ηp2 = .081

ECG Complex, but multiple

paths for analysis allow

for greater detail

25 mins ηp2 = .071 - .102

EMG Relatively

interpretable; could be

aided by phasic

analysis

25 - 30 mins ηp2= .279

EEG Complex; in this

research, some intuitive

and some non-intuitive

results

60 mins ηp2 = .137 - .389

* includes set-up, treatment, and analysis

209


interpretability of EDA, as well as its sensitivity to psychological stimuli, is reflected in its

widespread and stable rate of employment in psychophysiological and psychological research

(Dawson et al., 2000). This high rate of adoption is echoed in player experience and games

research, with the majority of studies reviewed in section 2 utilising EDA. The prevalent

employment of EDA in research literature represents a considerable advantage to researchers

considering a psychophysiological approach, particularly those unfamiliar with the

psychophysiological assessment. However, EDA produced one of the smallest effect sizes within

this program of research, possibly indicating limitations in the use of its assessment of small

differences between experiences; as such, despite its efficiency and robustness, EDA does not

necessarily allow for granular psychophysiological insight in the context of challenge-skill

balance manipulation. It should be noted that this conclusion should not be broadly applied to the

study of all concepts: as discussed in section 5.6, manipulation of separate phenomena—for

example, violence or sound—may yield different results, such as improved granularity.

Similar degrees of efficiency and accessibility are found for ECG (as EDA); however, this

is moderated somewhat by application and analysis. ECG is necessarily more invasive than palmar

EDA, requiring application to the torso of the participant—which can in turn amplify potential for

data loss in instances of increased adipose tissue, or the risk of dislodgement of electrodes

unknown to researcher or participant as a consequence of clothing obscuring the recording sites.

Despite this, ECG featured the second lowest rate of data loss within this program of research. In

terms of interpretability, the seemingly contradictory results of both HR and HRV HF Peaks

reported in this thesis suggest caution in the intuitive interpretation of results. While not unique to

ECG, core psychophysiological constructs—such as the startle response, anticipatory response,

directional fractionation, and habituation—are prone to influencing results, and warrant careful

interpretation and analysis of findings. Fortunately, multiple paths for ECG analysis (e.g. HR, R-

210


R, and HRV analysis) allow the opportunity for a granular and considered analysis approach. As

such, ECG situates itself as one of the more time-effective psychophysiological approaches while

simultaneously allowing for nuanced evaluation. Despite this, as with EDA, ECG featured some

of the lowest effect sizes reported within this program of research—once again indicating

limitations in the detection of small differences between experiences within the context of

challenge-skill balance manipulation. ECG may prove most useful for researchers interested in a

simultaneously efficient and robust psychophysiological assessment, notwithstanding possible

complexities associated with the interpretation of results.

With respect to EMG (taken from the OO site), the strength of the effect size, as well as

the interpretability of results, point to its strengths as a granular measure of the player experience

in the context of challenge-skill balance manipulation. Although EMG is typically used for phasic

analysis (through the study of reactive expressions to specific events), these results also establish

the usefulness of EMG in tonic analysis; however, phasic analysis would introduce greater clarity

to these results through the investigation of a possible startle response. Despite this, the EMG

measurement featured the greatest loss of data within this program of research—which may also

account for the absence of results altogether for EMG CS. This loss of data was likely the result

of poor electrode contact or electrode dislodgement at some point throughout the experience, and

a failure to recognise realtime data loss during experiment runtime. This issue may have been

circumvented through the use of more rigid equipment (during data collection it was noted that,

during setup, the ADD204 adhesive collars would often peel or crinkle on the surface of the skin

during moments of expressivity) and familiarisation with the appearance of data contamination in

real-time. Should these issues be resolved, however, EMG still proves costly in terms of temporal

investment: in setup, researchers need to be rigorous in ensuring they are accurately locating the

precise muscle site for analysis and may need to continually refer to a facial muscle chart

211


throughout experimentation to guarantee correct placement. Throughout data collection within

this program of research, both EMG sites were more prone than other measures to rates of greater

impedance on first-time application; this impedance was often high enough to warrant re-

preparation and re-application of the EMG electrodes. Finally, the distinction between movement

artefact and facial expressivity proved to be a time-consuming process in data treatment. As such,

the use of EMG is not readily recommended for time-poor studies or researchers not previously

familiar with psychophysiological assessment.

While results for EEG presented one of the highest effect sizes in the program of research,

outstripping all physiological and most subjective measures, the interpretability of results draws

into question the suitability of EEG—or perhaps more precisely, the EMOTIV Epoc—for player

experience research. Furthermore, while the application of the headset proved one of the least

time-consuming aspects of psychophysiological setup, the time cost and hardware resources

associated with frequency band analysis and data treatment offset this (particularly as each

frequency band for each site represents its own unique physiological measure, requiring individual

treatment and preparation). While the results EEG offers are rich and detailed, the associated time

costs, complexity of analysis and interpretation, and limitations of the EMOTIV Epoc (see section

5.5.6.4) situate this measure as potentially too restrictive in player experience research or

playtesting contexts.

Overall, the findings from the current program of research confirm that

psychophysiological assessment allows for objective and interesting insights otherwise

unattainable by typical evaluative measures within games research and evaluation (e.g. survey,

observation); however, the use of psychophysiology should be carefully considered in the context

of study aims and resources available to the research team. As the capability for distinguishing

212


small differences between experiences increase, so too does the complexities and resource-cost

associated with the psychophysiological measurement.

6.4.3 Sample Sizes

As discussed in section 2.9, prior player experience research employing multiple

contemporaneous psychophysiological measures has often featured small sample sizes. This

limitation emerged as a consequence of the substantial temporal investment required for the

attachment, treatment and evaluation of psychophysiological measures.

The question of sample size was addressed by this research program in two distinct ways:

as the conclusions suggest that psychophysiological assessment is limited in its capacity to predict

subjective experience in the context of challenge-skill balance manipulation, the evaluation of

granular differences may only be reliably achieved by obtaining a much larger sample size than

that reported in Study 2. This pushes the bounds of feasibility within current psychophysiological

research. However, commercial research has previously investigated the integration of non-

invasive physiological measures with video games hardware (such as controllers); this would

allow for a passive, remote assessment of psychophysiological response, which may generate the

sample size required for the psychophysiological distinction between more granular constructs.

The understanding of psychophysiological assessment as best suited for measuring large

differences also provides evidence in favour of sample sizes that do not necessarily exceed what

is required for assessing differences. It should be emphasised, however, that the limitations in

detecting small differences has only been established within this research in the context of

challenge-skill balance manipulation; future research may find greater success in the

psychophysiological detection of granular experiences.

213


6.4.4 Data Quality Checks

The notable loss of data for both EMG OO and EMG CS establishes a clear need for

adapting psychophysiological methodologies to include consistent psychophysiological analysis

and quality checks throughout data collection. Furthermore, where possible, the visual

familiarisation among researchers in the player experience space with a compromised or unclear

physiological signal would also aid in the early detection of unusable data. It is thus proposed that

physiological data should be treated and analysed every five n (dependent on intended sample

size); extra care is recommended for the collection of facial EMG data.

6.4.5 Automation and Reduction of Participant‒Researcher Interaction

One of the key strengths of this research program lay in automating the experimental

process and reducing participant‒researcher interaction, which has been revealed as problematic

in prior psychophysiological player experience research (Mandryk et al., 2006b). The Sequencer

software notably reduced the risk of human error in the experimental process, ensuring that the

appropriate condition and questionnaire order was maintained throughout and preserving

consistent runtime of all baseline and play conditions. Furthermore, the Sequencer software

allowed the researcher to remain physically removed from participants, with all interaction limited

to short instructions throughout (e.g., ‘Please press continue.’). This approach notably minimised

the potential for human interaction to influence physiological response, and is strongly

recommended for ongoing psychophysiological research in the player experience space. Although

not feasible for this research program, live recording of the participant’s screen would allow for

the researcher to be removed altogether to a separate room once data collection had commenced.

214


6.4.6 Familiarisation with Psychophysiological Principles

Player experience research should remain aware of fundamental psychophysiological

concepts in the evaluation of physiological data. Concepts such as habituation, startle response,

stimulus-response specificity, directional fractionation, one-to-many and many-to-one domain

relationships, and the myth of the unidimensional continuum of arousal should moderate the

analysis and interpretation of psychophysiological response. As discussed in 2.9.6, studies in the

player experience research space are sometimes limited in their consideration of these concepts as

potentially influential on, or responsible for, the presented results; therefore, it is advisable to

moderate conclusions with the general principles of psychophysiological interpretation.

6.4.7 Summary

Psychophysiological assessment allows for a quantitative, objective and insightful

evaluation of the player experience (Mandryk et al., 2006b; Bernhaupt et al., 2008). Using

psychophysiological measures may allow for experiences not detected through subjective

measures to be identified. In this research program, increased HR in the Boredom condition

suggests that the experience—despite largely being reported as sub-optimal, or boring, by

players—may also be influenced by anticipatory response. Additionally, psychophysiological

analysis also allows for an additional dimension in the interpretation of subjective results—while

players generally reported the Overload condition as sub-optimal in terms of interest/enjoyment,

presence, competence, autonomy and flow, increased EDA and decreased HF peaks in this

condition position it as a high-arousal experience. The use of psychophysiological measures in

conjunction with subjective measures allows for a more nuanced understanding of the player

experience.

215


The tonic interpretation of psychophysiological response should be mediated by the

consideration of stimulus-response specificity, habituation and domain relationships—for

example, the many-to-one relationship. While employing subjective measures may help to clarify

these experiences, care should be taken in interpreting physiological response as indicating certain

psychological states. This is further highlighted by the limitations in assessing predictive

relationships between psychophysiological response and specific self-reported components of the

player experience.

Overall, psychophysiological assessment allows for a greater understanding of the player

experience than that obtained through subjective evaluation alone. However, as the direct

psychological stimulus for physiological response is rarely conclusive, care should be taken in

tonic psychophysiological evaluation—n approach employing both psychophysiological and self-

report measures allows for a clearer and more comprehensive interpretation of the player

experience.

6.5 LIMITATIONS AND FUTURE RESEARCH

This research has provided insight into the psychophysiological response to optimal and

sub-optimal play experiences, psychophysiological methodologies within the player experience

evaluation space and the applicability of psychophysiological measures in evaluating the player

experience within academic and commercial contexts. However, as revealed by analysis and

discussion, there are methodological limitations that leave some components unexplored.

As the Overload condition was designed to overwhelm, frustrate or evoke anxiety in

participants, a semi-counterbalanced design was introduced to mitigate the potential for a long-

lasting negative mood induction inadvertently influencing the remaining conditions. Within the

216


semi-counterbalanced design, the Overload condition was placed to always occur last in condition

order; both the Boredom and Balance condition were fully counterbalanced. Consequently,

potential exists for order effect to influence the player experience data collected from the Overload

condition. One potential impact of the semi-counterbalanced approach may be the possibility for

physiological drift to influence EDA results. As EDA presents an additive signal, it is possible for

the signal to continually increase overtime over the course of the experiment; as such, this could

potentially influence the increased EDA signals found in the Overload condition. However,

Braithwaite et al. (2015) note that drift is primarily only an issue in long-term experiments, and

especially long-term ambulatory experiments; furthermore, the impact of potential drift has been

somewhat mitigated by the inclusion of baselines between each play session. Despite this, future

research may entail further data collection within a fully counter-balanced version of the current

study methodology, as well as the use of the baseline data as a calibration point for EDA analysis

and investigation of physiological drift.

Additionally, the Overload condition was the only condition in which player death was

unavoidable. As each player death represented up to 15 seconds of removal from play (in which

the screen would fade, and reset the player to the beginning area), disengagement from the play

experience during this period may have occurred. While player deaths were also possible in the

Balance condition, they were not ensured by the paradigms of the condition design (and were

impossible in the Boredom condition). The increased death rate in the Overload condition may

thus have similarly increased the risk of disengagement.

Due to limitations in the time available for analysis within the candidature and

consideration for applicability to player experience evaluation contexts, physiological analysis

was restricted to tonic evaluation. Mandryk et al. (2006b) identify the prohibitive temporal costs

217


associated with phasic analysis and associated video coding in their own studies of eight and 10

participants; as this study featured an n of 90, tonic analysis was chosen as the most suitable

method for physiological evaluation. However, as discussed in section 5.5.4, this does have

implications for the assessment of EMG—a measure that is typically evaluated phasically, and

does not consistently allow for post-hoc distinction between facial movements. Phasic analysis of

the available data, particularly EMG, will be considered in further exploration of the program’s

results moving forward. Future research may also benefit from an aggregate approach, as used by

Hazlett (2006), in which Hazlett analysed the physiological response to video game events that

were previously identified by expert participants as either ‘negative’ or ‘positive’

An additional analysis approach that may be insightful in terms of explaining results is a

comparison of physiological timebins. This is especially applicable to physiological analysis of

the Overload condition, in which habituation or physiological detachment were identified as

potentially influential on HR data. In particular, HR analysis and comparison of the first and last

five minutes of the Overload would allow for a more complete understanding of the

psychophysiological response to the condition.

The analysis of the EEG sites was also limited as a result of research scope considerations.

While the EMOTIV Epoc EEG headset provides 14 channels (each recording from a specific site),

only two channels—or two sites—were selected for frequency analysis. Furthermore, analysis was

restricted to the alpha, beta and theta frequency bands. Complete analysis of all sites and frequency

bands (alpha, beta, gamma, theta and delta) would have resulted in the generation of 70 variables

for inclusion in the analyses described in section 5.1.8. Due to limitations in laboratory equipment,

each site and frequency band required manual scanning to detect and remove movement artefacts.

Analysis was thus restricted to two sites located on the right hemisphere, and three frequency

218


bands (or six variables), chosen for reasons described in section 3.3.3.1. Depending on the

determination of the EMOTIV Epoc EEG headset’s usefulness in this research program, the

analysis of these additional sites and frequency bands presents an opportunity for a more granular

understanding of the electroencephalographic response to player experience.

Unfortunately, a considerable amount of data was lost from both the OO and CS EMG

sites. This reduced sample size may have been responsible for the absence of results discovered

for the CS site in particular. The loss of data was likely a result of poor electrode contact, electrodes

shifting throughout the experiment or a fault in one of the electrodes in circulation; an alternative,

although less likely, explanation is signal interference at the laboratory site. Regrettably, the poor

data quality was unrealised during collection due to lack of visual familiarity with compromised

EMG signals; this issue was thus not discovered during occasional exploratory analysis throughout

data collection. Thorough analysis and checking of physiological data, as well as familiarisation

with compromised signals, is recommended throughout data collection in future research. As

EMG proved especially sensitive to noise and data loss within this research program, special care

is advised for employment of this measure.

Additional physiological measures not employed in this research program may also help

to shed light on the current understanding of psychophysiological response to player experience.

Omitted from Study 2’s methodology for temporal, financial or practical reasons detailed in

section 3.3.3 are measures that include BP, respiration, salivary cortisol, eye-tracking, body

temperature analysis and the ZM site for EMG. Including these measures in future research would

enable a more granular understanding of the psychophysiological player experience.

In terms of participant demographics, female participants were notably underrepresented,

with 22.5% female participation in Study 2; this contrasts with data available on video game player

219


demographics, indicating that between 44% (Entertainment Software Association, 2015) and 47%

(Brand & Todhunter, 2015) of video game players are female. It is unknown whether the gender

ratio for Study 2 affected the generalisability of results. This gender imbalance points to limitations

in sampling from undergraduate STEM cohorts, in which women are disproportionately

underrepresented (Hill, 2010). As under-employment of female participants has been identified as

a consistent limitation in player experience research (Järvelä et al., 2015), there is a recognisable

need to address gender representation in future research in this space.

Gender likewise represents a potential limitation in the current interpretation of results.

Some research suggests that gender does play a role in, and may influence, psychophysiological

results; consequently, many psychophysiological studies are performed using a male sample only

(Stern, 2001). Bianchin & Angrilli (2012) suggest biological grounding for greater sensitivity and

vulnerability to adverse or stressful events amongst women, manifesting in increased startle reflex

amplitude and modulation in the amygdala and orbitofrontal cortex. In their investigation of

children’s psychophysiological response to violent video games, Gentile et al. (2017) found

increased cardiovascular activity in response to the violent condition amongst boys only. Despite

this, Gentile et al. note that similar studies have not returned this gender disparity in results, and

so suggests further investigation. Future analysis of the results reported within this thesis would

benefit from separate treatment of data by gender, for the dual benefits of investigating a potential

gender influence on results and contributing to current understanding of the role of gender in

psychophysiological measurement and analysis.

As for the generalisability of results, the video game conditions used for this research

program were restricted to a single-player first-person shooter PC game. Future studies in this

space may benefit from expanding this research to include additional genres, environments and

220


platforms in the interest of results’ generalisability. Furthermore, as human interaction has been

found to have a profound effect on physiological response (Mandryk et al., 2006b), future research

may benefit from a large-scale psychophysiological evaluation of a social play experience.

The difficulties of assessing flow in an easy or unchallenging play experience point to a

clear need for continued assessment of the construct, as well as the S FSS and FSS-2 flow scales,

in terms of applicability to player experience evaluation. As the S FSS and FSS-2 are closely

matched to Csikszentmihalyi’s (1990) criteria for the experience of flow, there may be merit in

considering whether these criteria can be directly applied to evaluating the video game player

experience. This is particularly pertinent due to the prominence of the flow construct in player

experience literature (Cowley et al., 2008). Furthermore, continued exploration of the role of

challenge‒skill balance in evoking flow is recommended; the results of this research potentially

further support recent research disputing the concept of challenge‒skill balance as an antecedent

to flow (Fong et al., 2015; Rheinberg et al., 2003). Similarly, the assessment of competence within

the PENS scale yielded comparable difficulties in distinguishing between boring and optimal play

experiences. This suggests caution in interpreting high competence as reported by the PENS scale

as indicating optimal play experience. This route of investigation should also be applied more

generally to the assessment of unchallenging games or optimal play experiences, with some

conclusions drawn from this research suggesting the possibility for players to invent their own

‘fun’ or challenge within inherently un-fun play experiences.

The potential for player-created challenge may be amplified within experimental settings,

as the option to simply quit playing as a result of boredom is not always available. Mandryk et al.

(2006b) note a similar phenomenon in the easy condition generated for their study on

psychophysiological techniques; the authors recount players creating their own challenge within

221


NHL 2003 by introducing personal rules such as scoring a goal with only defensemen. This is also

true for overly-challenging experiences: in an experimental setting, the player is obligated to

continue playing or may feel under social pressure to self-moderate negative response to difficult

play experiences. As such, the Overload and Boredom conditions may not be completely

generalisable to a typical player experience due to their experimental context. Consequently, the

artificiality of the laboratory environment is another limitation of this program of research (as well

as most extant player experience literature). Future psychophysiological player experience

research may benefit from the employment of naturalised experimental settings that encourage

autonomous interactions with play conditions. Another option may be the remote collection of

physiological data via wireless worn devices, so that participants may play the video game

conditions at home.

Finally, the data obtained during this research program did not allow for the

psychophysiological identification of small changes in specific components of the player

experience. A combination of data loss from some psychophysiological measures, as well as the

low variances for HF, EMG OO and EEG, resulted in a dearth of results for both correlations and

regressions analysis. This may suggest that, in terms of predictive analysis, psychophysiological

measures may be limited in their detection of small differences in play experiences in the context

of challenge-skill balance manipulation. It should be noted that this is not currently generalisable

to the study of other player experience phenomena; future research of other video game concepts

and traits, such as the manipulation of sound or graphical violence, should incorporate predictive

analysis to determine whether this result is also true within the context of those specific

manipulations. However, within the context of challenge-skill balance manipulation, the

limitation in the detection of small differences may be ameliorated through the collection of a

larger sample size. While the collection of a notably larger sample size than that featured in this

222


research is not feasible within most academic and commercial contexts, including non-invasive

sensors for collecting physiological data in games hardware (e.g., controllers) would allow for the

passive collection of psychophysiological data from a far larger sample pool. Research in this

space has already been attempted by video game development companies Nintendo, Sony and

Valve (Patel, 2009; Findlater, 2013; Ambinder, 2011). The passive obtainment of a substantial

sample size would generate a more refined understanding of the psychophysiological responses to

play.

6.5.1 Future Approaches to Analysis

A tonic approach was selected for this program of research in alignment with extant player

experience literature, and out of consideration of time constraints; however, future interpretation

of the data presented within this thesis may also benefit from more granular or phasic analysis.

This section will detail several approaches that could be undertaken that will allow for further

exploration of the data collected within this program of research, and that may grant further clarity

to the current interpretation of the results.

Further investigation of the unexpected EEG O2 Alpha and HR results may benefit from a

more phasic analysis in the interpretation of initial and concluding timebins. In particular, a

frequency analysis of a select period (e.g. 3—5 minutes) from near the start of each condition and

from the end of each condition may reveal a pattern of results indicating habituation, or

anticipatory response, as a possible explanation for the results. For example, significantly

increased HR in the Boredom condition may only emerge as a result for the initial timebin—which

would further support the conclusion of the role of anticipatory response in the interpretation of

this data. This analysis approach could also contribute to an understanding of the player experience

as it changes over time. Participant recollection of experiences is often influenced by a peak-end

223


effect, in which the ‘peak’ (or climax) and the conclusion of the experience colours the subjective

response to the experience (Gutwin, Rooke, Cockburn, Mandryk, & Lafreniere, 2016); as such,

real-time psychophysiological exploration of the experience as it differs over time may present

novel findings that improve current understanding of player experiences.

A phasic analysis of events may also present the opportunity for clarification of the findings

drawn from the tonic analysis; however, it is important to note that neither study within this

program of research was designed for phasic analysis. As such, all play conditions lack the discrete

events that typically occur in video game artefacts intended for phasic analysis—such as picking

up tokens, or falling off the map, in the phasic psychophysiological research undertaken by Ravaja

et al. (2006). Furthermore, as challenge and challenge-skill balance were a primary focus of this

research, the introduction of DDA to the Balance condition resulted in differing combat

experiences per participant in the interest of maintaining consistent challenge (e.g. increased or

fewer special enemies, increased or decreased combat duration). Despite this, one potential path

for phasic analysis is that of the deaths that occur in the Overload condition: analysis of

physiological response to repeated player deaths may help to explore conclusions offered by

Ravaja et al. in earlier research (2008).

Normalisation of data represents an additional route for further analysis of the data collected

within this program of research. Mandryk (2008) recommends the normalisation of physiological

data – or representation of physiological data as a percentage of the timespan – for improved

contextualisation and interpretation, and reduction of the impact of individual physiological

differences (Mandryk, 2008). It is possible that normalising the collected data may allow for some

patterns to emerge when investigating individual correlations that were not otherwise accessible

224


without applying normalisation techniques. As such, normalisation represents a natural future step

in analysis.

Further research could include more detailed analysis through the interpretation of clusters

of data. One route for exploring the data obtained during this research program is to consider

possible differences between the psychophysiological responses of novice and expert players,

which was not assessed here. Research in the psychophysiological space has identified expertise

as having influence on the physiological response (Fairclough, Venables, & Tattersall, 2004;

Cooke et al., 2014), with evaluation of the player experience having more specifically identified

associations between expertise in video game play and decreased emotional expressivity as

recorded by EMG (Weinreich, Strobach, & Schubert, 2015). A large-scale psychophysiological

comparison of novice and expert groups within video game play would broaden current

understanding of the psychophysiological response to player experience. Other options for

exploration may include age, participant familiarity with the chosen game artefact and participant

familiarity with genre.

6.6 CONTRIBUTIONS TO KNOWLEDGE

This program of research has explored the utility of the psychophysiological assessment of

the player experience and has contributed recommendations for ongoing psychophysiological

approaches in terms of the selection of physiological measures, experimental design, analysis, and

data collection. The findings driven by both the subjective and psychophysiological data have

further contributed to extant psychophysiological player experience literature in the exploration

of relationships between challenge and physiological response, finding generally increased

physiological activation in instances of high challenge (with some exceptions). These findings

have both supported and contrasted with existing research, and consideration—based on core

225


psychophysiological concepts—has been given to why this might be. This has notably fulfilled a

component of the overarching aim for this program of research, which sought to ‘… further clarify

existing contributions to literature’.

The finding of generally increased physiological activation in instances of high challenge

presents several implications for both industry and research. In the development of biofeedback

and physiologically-informed DDA, researchers and developers will find greater assurance in the

assumption that increased challenge will result in increased arousal; as such, there is improved

credence to the concept of the dynamic adjustment of difficulty in response to decreases and

increases of arousal. In terms of player experience assessment, researchers and developers may

benefit from this knowledge when assessing the difficulty of their game. Despite this, as suggested

by some results in this program of research, it is pertinent that researchers and developers remain

aware of the potential for habituation – and that a sudden decrease in arousal may not necessarily

indicate a decrease in challenge.

Relatedly, this thesis contributes several methodological recommendations for

psychophysiological analysis within the field of player experience evaluation. Chief among these

is the careful consideration of core psychophysiological principles, such as habituation, stimulus-

response specificity, and directional fractionation, which have often been neglected in results

interpretation in previous player experience literature. It is hoped that the discussion of these

concepts, and how they may influence results, will allow for improvements in the careful

consideration of psychophysiological results in future player experience research.

Furthermore, the identification of the benefits and drawbacks of the psychophysiological

measures used within this study may help in the development of future player experience

methodologies. In particular, consideration of time costs—in terms of both deployment and

226


analysis—may be of notable importance to researchers, not least because of the unique

requirement in player experience research to limit the potential for participant disengagement or

boredom. Additionally, while the results of this research aren’t necessarily generalisable, the

performance of the psychophysiological measurements reported within may also be helpful to

future researchers developing an experiment methodology: for example, EDA was found to be

simultaneously the most time-efficient measure while also having the lowest effect size; in

contrast, EEG was the most time-consuming measure while also featuring the greatest effect size

(amongst some frequency bands). While this can vary by equipment, expertise, analysis approach,

and subject matter, and is therefore not universal, these findings present a helpful starting point

for methodology development and consideration.

As discussed in section 6.4.3, this research also highlights the unique benefits that industry

and industrial research possess: the opportunity for the remote, naturalised, and widespread

collection of data, allowing for the large sample size required for the psychophysiological

distinction between granular constructs. The analysis of such data would notably improve

understanding of player bases, allow unique insights for developers in terms of creating optimal

play experience, and represent a substantial contribution to player experience literature. The

potential contributions of this work extend beyond player experience, and carry implications for

broader HCI and psychology research. As growing literature employs games as a stimulus for

studying human behaviour (Järvelä et al., 2015), there is greater emphasis on understanding the

fundamental artificiality of the laboratory and movement towards richer stimulation paradigms

that are closer to real world experiences. As such, the contained consideration and review of a

psychophysiological method employing a video game as the core stimulus may prove beneficial

for research for which video game design and player experience analysis is not the research

objective. Similarly, challenge is not a construct unique to player experiences—a natural analogue

227


is challenge that occurs in learning or vocational environments. As such, contributions to the

psychophysiological understanding of challenge may also be useful to research investigating the

effects of differing challenge levels, or over- and underwhelming challenge, on students and

workers. Finally, the consideration of the utility of psychophysiological assessment—and

recommendations for best practice—is useful to any researcher who may be considering a

psychophysiological approach in their own work, regardless of academic or commercial contexts.

6.7 CONCLUSION

This research expanded upon the understanding of psychophysiological experience of

play, as well as the utility of physiological response as a means of assessing the player experience.

Stage 1 (Chapter 3) laid the foundations of the research program by exploring a viable

psychophysiological and psychological approach for the analysis of player experience. Review of

player experience and psychophysiological literature identified EMG, ECG, EDA and EEG, as

well as the subjective experience of flow, as the initial measures that would allow for a robust

exploration of the psychophysiological response to player experiences. An iterative design stage

resulted in the creation of three video game conditions—Boredom, Balance and Overload—that

differed in terms of challenge‒skill balance, providing a useful tool to compare psychological and

psychophysiological responses.

The experimental study conducted in Stage 2 (Chapter 4) investigated the three video

game conditions’ ability to invoke or inhibit flow, and found that results did not conform to

expectations of condition performance. Namely, flow was not found to differ significantly

between the Boredom and Balance conditions. A robust investigation of these results presented

several possible explanations, as discussed in Chapter 4.

228


In response to these findings, several changes were made to the study methodology and

condition design. The psychometric approach was adapted to include additional measures

(interest/enjoyment, competence, autonomy and presence), with the assessment of flow restricted

to a shorter scale to minimise experiment runtime. The challenge‒skill condition manipulation

was maintained, as challenge remained a crucial element in ensuring optimal play experiences.

The Boredom condition was further modified to ensure the absence of game challenge, with all

semblance of combat—in the form of enemy combatants—removed from the game. Furthermore,

Stage 2 oversaw the development of the Sequencer and baseline experimental software to improve

the rigour of physiological analysis. Both programs were designed to minimise participant‒

researcher interaction, ensure consistent experiment, baseline and condition runtimes, and

guarantee appropriate condition and survey order.

Stage 3 (Chapter 5) found that the condition design was generally successful in creating

optimal and sub-optimal conditions, with the Balance condition emerging as the intended optimal

player experience. In all self-reported measures, Balance received the highest of equal highest

results. For the flow and competence measures, Balance was matched by Boredom in this

achievement, but was at no point outstripped by Overload. These results support and expand upon

the crucial role of challenge and challenge‒skill balance in creating optimal player experiences.

However, interesting implications for assessing sub-optimal conditions—especially conditions in

which the skill of the player radically outstrips the demands of the game—also arise. These

findings suggest the potential for player-generated challenge fun in boring or underwhelming

conditions. Not exclusive from this conclusion is also possible limitations of the PENS

competence and the Jackson FSS-2 and S FSS-2 in assessing boring or easy video game artefacts,

as the items often do not allow for a clear distinction between challenge‒-skill balance and skill >

challenge imbalance experiences.

229


Analysis of physiological response expands on the current understanding of the

psychophysiological response to player experiences, and reveals results both congruent and

incongruent with current player experience research. Results for the Overload condition reveal

significantly increased EDA and decreased HRV HF peaks than in the Boredom or Balance

conditions; furthermore, they featured significantly increased HR than the Overload condition. As

previous psychophysiological research in the player experience space found no main effect of

difficulty on EDA, HR or HRV, these results represent a novel contribution to the field. EMG OO

activity was found to be significantly greater in the Overload condition than either the Boredom

or Balance conditions, tentatively supporting previous player experience research that found

decreases in OO activity associated with boring player experiences. This supports EMG OO as a

potentially reliable measure of experienced difficulty. Finally, the findings for the EEG frequency

bands may suggest states of greater mental and emotional processing, attentional focus, creativity

and potentially negative affect in the Balance and Overload conditions than in the Boredom

condition, linking these cognitive states with increased challenge (or suggesting the inhibition of

such states in boring or unchallenging play experiences).

The analysis and discussion of the data gathered during Stage 3 also allowed for the evaluation

of the successes and limitations of the methodology employed for this research program.

Furthermore, applying general psychophysiological concepts—such as stimulus-response

specificity, directional fractionation or the consideration of one-to-many domain relationships—

facilitated analysis informed by psychophysiological practice and literature. These findings

enabled the generation of a series of recommendations, cautions and opportunities for future

psychophysiological research in the player experience space (discussed in sections 6.4 and 6.5),

particularly regarding practical concerns (such as time costs) and recommended caution in

interpretation. Generally, the analysis of psychophysiological response was found to be useful for

230


generating insightful, objective conclusions regarding the player experience; however, these

results are tempered by the limitations of the inconclusive nature of physiological response, and

recommendations are made for methodological approaches featuring both psychophysiological

and subjective measures.

6.8 FINAL COMMENTS

As video games continue to grow as a leading form of entertainment, so too does the value

of understanding the player experience. The use of psychophysiological assessment in a context

typically evaluated by subjective and self-report measures allows for a broader understanding of

the player experience, and enables exclusive objective insights that complement subjective

evaluation. This program of research represented an effort to further expand psychophysiological

understanding of the player experience through a large-scale approach, featuring a larger sample

size and greater breadth of physiological measures than was currently available in player

experience literature. This thesis presents novel contributions in the form of psychophysiological

relationships with challenge and challenge‒skill balance, an important element of optimal play

experiences. Furthermore, this thesis highlights the challenges and benefits of

psychophysiological analysis, and contributes methodological and interpretative

recommendations for future psychophysiological evaluation in player experience research.

231


7 REFERENCES

Abuhamdeh, S., & Csikszentmihalyi, M. (2012). The Importance of Challenge for the Enjoyment

of Intrinsically Motivated, Goal-Directed Activities. Personality and Social Psychology

Bulletin, 38(3), 317‒330. doi: 10.1177/0146167211427147

Alexander, J. T., Sear, J., & Oikonomou, A. (2013). An investigation of the effects of game

difficulty on player enjoyment. Entertainment Computing, 4(1), 53–62. doi:

10.1016/j.entcom.2012.09.001

Allen, M. T., Obrist, P. A., Sherwood, A., & Crowell, M. D. (1987). Evaluation of myocardial and

peripheral vascular responses during reaction time, mental arithmetic, and cold pressor

tasks. Psychophysiology, 24(6), 648‒656. doi: 10.1111/j.1469-8986.1987.tb00345.x

Ambinder, M. (2011). Biofeedback in Gameplay: How Valve Measures Physiology to Enhance

Gaming Experience. Poster session represented at Information Online. Retrieved from

http://www.information-

online.com.au/sb_clients/iog/data/content_item_files/000001/PresentationC1.pdf

Andrade, G., Ramalho, G., Gomes, A. S., & Corruble, V. (2006). Dynamic game balancing: An

evaluation of user satisfaction. The AAAI Press.

Andreassi, J. L. (2007). Psychophysiology: Human Behavior & Physiological Response (5th ed.).

Mahwah, NJ: Lawrence Erlbaum Associates.

Baldwin, A., Johnson, D., Wyeth, P., & Sweetser, P. (2013). A framework of Dynamic Difficulty

Adjustment in competitive multiplayer video games. IEEE Consumer Electronics

Society’s International Games Innovations Conference, IGIC, 16‒19. doi:

10.1109/IGIC.2013.6659150

Baldwin, A., Johnson, D., & Wyeth, P. (2014). The effect of multiplayer dynamic difficulty

adjustment on the player experience of video games. Proceedings of the 2014 CHI


1489‒1494, Toronto, ON. doi: 10.1145/2559206.2581285

Baldwin, A. (2017). Balancing act: the effect of dynamic difficulty adjustment in competitive

multiplayer video games (Doctoral thesis). Retrieved from

https://eprints.qut.edu.au/102669/

Ballard, M. E., & West, J. R. (1996). Mortal Kombat: The effect of violent videogame play on

males’ hostility and cardiovascular responding. Applied Social Psychology, 26(8), 717‒

730. doi: 10.1111/j.1559-1816.1996.tb02740.x

Baumeister, J., Barthel, T., Geiss, K. R., & Weiss, M. (2008). Influence of phosphatidylserine on

cognitive performance and cortical activity after induced stress. Nutritional

Neuroscience, 11(3), 103‒110. doi: 10.1179/147683008X301478

Bazanova, O. M., & Vernon, D. (2014). Interpreting EEG alpha activity. Neuroscience and

Biobehavioral Reviews, 44, 1-17. doi: 10.1016/j.neubiorev.2013.05.007

Bernhaupt, R., Ijsselsteijn, W., Mueller, F., Tscheligi, M., & Wixon, D. (2008). Evaluating user

experiences in games. CHI ’08 Proceedings, 3905–3908, Florence, Italy. doi:

10.1145/1358628.1358953

Berntson, G. G., Bigger, J. T., Jr., Eckberg, D. L., Grossman, P., Kaufmann, P. G., Malik, M., …

https://doi.org/10.1145/1358628.1358953

232


van der Molen, M. W. (1997). Heart rate variability: Origins, methods, and interpretive

caveats. Psychophysiology, 34(6), 623‒648. doi: 10.1111/j.1469-8986.1997.tb02140.x

Bianchin, M., & Angrilli, A. (2012). Gender differences in emotional responses: A

psychophysiological study. Physiology & Behaviour, 105(4), 925-932. doi:

10.1016/j.physbeh.2011.10.031

Billman, G. E. (2013). The LF/HF Ratio Does Not Accurately Measure Cardiac Sympatho-vagal

Balance. Frontiers in Physiology, 4(26), 1‒12. doi: 10.3389/fphys.2013.00026

Biocca, F., Harms, C., & Burgoon, J. K. (2003). Toward a more robust theory and measure of

social presence: Review and suggested criteria. Presence: Teleoperators and Virtual

Environments, 12(5), 456‒480. doi: 10.1162/105474603322761270

Boehner, K., DePaula, R., Dourish, P., & Sengers, P. (2007). How emotion is made and measured.

Intl. Journal of Human-Computer Studies, 65(4), 275-291. doi:

10.1016/j.ijhcs.2006.11.016

Boucsein, W. (1992). Electrodermal Activity. New York, NY: Plenum.

Boyle, E. A., Connolly, T. M., Hainey, T. & Boyle, J. M. (2012). Engagement in digital

entertainment games: A systematic review. Computers in Human Behavior, 28(3), 771‒

780. doi: doi: 10.1016/j.chb.2011.11.020

Bradley, M. M. (2000). Emotion and Motivation. In J. T. Cacioppo, L. G. Tassinary, & G. G.

Berntson (Eds.), Handbook of Psychophysiology (2nd ed.) (pp. 602‒642). New York, NY:

Cambridge University Press.

Braithwaite, J. J., Watson, D. G., Jones, R., & Rowe, M. (2015). A Guide for Analysing

Electrodermal Activity (EDA) & Skin Conductance Responses (SCRs) for Psychological

Experiments. Retrieved from http://www.birmingham.ac.uk/documents/college-

les/psych/saal/guide-electrodermal-activity.pdf

Brand, J. E., & Todhunter, S. (2015). Digital Australia Report 2016. Retrieved November, 2015,

from http://www.igea.net/wp-content/uploads/2015/07/Digital-Australia-2016-DA16-

Final.pdf

Brühlmann, F., & Schmid, G.-M. (2015). How to Measure the Game Experience?: Analysis of the

Factor Structure of Two Questionnaires. Proceedings of the 33rd Annual ACM


1181‒1186, Seoul, Republic of Korea. doi: 10.1145/2702613.2732831

Cacioppo, J. T., Tassinary, L. G., & Berntson, G. G. (2000). Handbook of Psychophysiology (2nd

ed.). New York, NY: Cambridge University Press.

Chatfield, T. (2010). Fun Inc.: Why Play is the 21st Century’s Most Serious Business. London,

England: Virgin Books.

Chen, J. (2007, April 1). Flow in games (and everything else). Communications of the ACM. doi:

10.1145/1232743.1232769

Cinaz, B., Marca, R., Arnrich, B., & Tröster, G. (2013). Monitoring of mental workloads levels

during an everyday life office-work scenario. Personal and Ubiquitous Computing, 17(2),

229‒239. doi: 10.1007/s00779-011-0466-1

Cole, H. W., & Ray, W. J. (1985). EEG correlates of emotional tasks related to attentional

demands. Intl. Journal of Psychophysiology, 3(1), 33‒41. doi: 10.1016/0167-

http://dx.doi.org/10.1162/105474603322761270

http://dx.doi.org/10.1016/j.chb.2011.11.020

http://doi.org/10.1145/1232743.1232769

http://doi.org/10.1145/1232743.1232769

233


8760(85)90017-0

Cole, T., Cairns, P., & Gillies, M. (2015). Emotional and Functional Challenge in Core and Avant-

garde Games. Proceedings of CHI Play ’15, 12‒126, London, UK. doi:

10.1145/2793107.2793147

Cooke, A., Kavussanu, M., Gallicchio, G., Willoughby, A., McIntyre, D., & Ring, C. (2014).

Preparation for action: psychophysiological activity preceding a motor skill as a function

of expertise, performance outcome, and psychological pressure. Psychophysiology, 51(4),

374‒384. doi: 10.1111/psyp.12182

Cote, A., & Raz, J. G. (2015). In-depth interviews for games research. In P. Lankoski & S. Björk

(Eds.), Game Research Methods (pp. 93‒116). ETC Press. Retrieved from

http://press.etc.cmu.edu/files/Game-Research-Methods_Lankoski-Bjork-etal-web.pdf

Coulson, M., Barnett, J., Ferguson, C. J., & Gould, R. L. (2012). Real feelings for virtual people:

Emotional attachments and interpersonal attraction in video games. Psychology of

Popular Media Culture, 1(3), 176‒184. doi: 10.1037/a0028192

Cowley, B., Charles, D., Black, M. & Hickey, R. (2008). Toward an understanding of flow in

video games. Computers in Entertainment, 6(2), 1‒27. doi: 10.1145/1371216.1371223

Cox, A., Cairns, P., Shah, P., & Carroll, M. (2012). Not Doing But Thinking: The Role of

Challenge in the Gaming Experience. Proceedings of the CHI 2012, 79–88. doi:

10.1145/2207676.2207689

Csikszentmihalyi, M. (1990). Flow: The Psychology of Optimal Experience. In H. Collins (Ed.),

Annals of Physics (Vol. 54). Harper & Row. doi: 10.1145/1077246.1077253

Csikszentmihalyi, M. (2000). Beyond boredom and anxiety. San Francisco: Jossey-Bass.

Davidson, R. J., Jackson, D. C., & Larson, C. L. (2000). Human Electroencephalography. In J. T.

Cacioppo, L. G. Tassinary, & G. G. Berntson (Eds.), Handbook of Psychophysiology (2nd

ed.) (pp. 27‒52). New York, NY: Cambridge University Press.

Dawson, M. E., Schell, A. M., & Filion, D. L. (2000). The Electrodermal System. In J. T.

Cacioppo, L. G. Tassinary, & G. G. Berntson (Eds.), Handbook of Psychophysiology (2nd

ed.) (pp. 200‒223). New York, NY: Cambridge University Press.

Deci, E. L., & Ryan, R. M. (1980). The Empirical Exploration of Intrinsic Motivational Processes.

In B. Leonard (Ed.), Advances in Experimental Social Psychology (pp. 39–80). Academic

Press. doi: 10.1016/s0065-2601(08)60130-6

Deci, E. L., & Ryan, R. M. (2000). The ‘what’ and ‘why’ of goal pursuits: Human needs and the

self-determination of behavior. Psychological Inquiry, 11(4), 227‒268. doi:

10.2307/1449618

Dimberg, U. (1990). Facial electromyography and emotional reactions. Psychophysiology, 27(5),

481‒494. doi: 10.1111/j.1469-8986.1990.tb01962.x

Drachen, A., Nacke, L. E., Yannakakis, G. N., & Pedersen, A. L. (2010). Correlation between

heart rate, electrodermal activity and player experience in first-person shooter games. …

on Video Games, 49–54, LA, California. doi: 10.1145/1836135.1836143

Duvinage, M., Castermans, T., Petieau, M., Hoellinger, T., Cheron, G., & Dutoit, T. Performance

of the Emotiv Epoc headset for P300-based applications. BioMedical Engineering

OnLine, 12(56). doi: 10.1186/1475-925X-12-56

http://dx.doi.org/10.1145/2793107.2793147

http://dx.doi.org/10.1145/2793107.2793147


http://dx.doi.org/10.1037/a0028192

http://dx.doi.org/10.1145/1371216.1371223

http://doi.org/10.1145/2207676.2207689

http://doi.org/10.1145/2207676.2207689

http://doi.org/10.1016/s0065-2601(08)60130-6

http://dx.doi.org/10.2307/1449618

http://dx.doi.org/10.2307/1449618

https://doi.org/10.1145/1836135.1836143

http://dx.doi.org/10.1186/1475-925X-12-56

234


Ekman, P., Davidson, R. J., & Friesen, W. V. (1990). The Duchenne smile: Emotional expression

and brain physiology II. Personality and Social Psychology, 58(2), 342‒353. doi:

10.1037/0022-3514.58.2.342

Ekman, P. (1992). An Argument for Basic Emotion. Cognition and Emotion, 6(3-4), 169-200. doi:

10.1080/02699939208411068

Entertainment Software Assotiation. (2015). 2015 Essential Facts About the Computer and Video

Game Industry. Social Science Computer Review, 4(1), 2–4. Retrieved from

http://www.theesa.com/facts/pdfs/ESA_EF_2008.pdf

Fairclough, S. H., Venables, L., & Tattersall, A. (2005). The influence of task demand on the

psychophysiological response. Intl. Journal of Psychophysiology, 56(2), 171‒184. doi:

10.1016/j.ijpsycho.2004.11.003

Fernández, T., Harmony, T., Rodríguez, M., Bernal, J., Silva, J., Reyes, A., & Marosi, E. (1995).

EEG activation during the performance of tasks involving different components of mental

calculation. Electroencephalography and Clinical Neurophysiology, 94(3), 175‒182. doi:

10.1016/0013-4694(94)00262-J

Findlater, W. (2013, August 26.) Mark Cerny: ‘The PlayStation 4 will recapture gaming’s glory

days.’ Stuff. Retrieved from http://www.stuff.tv/

Fleming, M. J., & Rickwood, D. J. (2001). Effects of violent versus nonviolent video games on

children’s arousal, aggressive mood, and positive mood. Applied Social Psychology,

31(10), 2047‒2071. doi: 10.1111/j.1559-1816.2001.tb00163.x

Fong, C. J., Zaleski, D. J., & Leach, J. K. (2015). The challenge‒skill balance and antecedents of

flow: A meta-analytic investigation. Journal of Positive Psychology, 10(5), 1‒22. doi:

10.1080/17439760.2014.967799

Fridlund, A. J., & Cacioppo, J. T. (1986). Guidelines for human electromyographic research.

Psychophysiology, 23(5), 567‒589. doi: 10.1111/j.1469-8986.1986.tb00676.x

Gentile, D.A., Bender, P. K., & Anderson, C. A. (2017). Violent video game effects on salivary

cortisol, arousal, and aggressive thoughts in children. Computers in Human Behavior, 70,

39-43, doi: 10.1016/j.chb.2016.12.045

Giakoumis, D., Tzovaras, D., Moustakas, K., & Hassapis, G. (2011). Automatic Recognition of

Boredom in Video Games Using Novel Biosignal Moment-Based Features. IEEE

Transactions on Affective Computing (TAC), 2(3), 119–133.

Harmat, L., de Manzano, Ö, Theorell, T., Högman, L., Fischer, H., & Ullén, F. (2015).

Physiological correlates of the flow experience during computer game playing.

International Journal of Psychophysiology, 97(1), 1‒7. doi:

10.1016/j.ijpsycho.2015.05.001

Hazlett, R. L. (2006). Measuring emotional valence during interactive experiences: boys at video

game play. In Proceedings of the SIGCHI Conference on Human Factors in Computing

Systems, 1023-1026, Montreal, Canada. doi: 10.1145/1124772.1124925

Hill, C., Corbett, C., & St. Rose, A. (2015). Why So Few? Women in Science, Technology,

Engineering, and Mathematics. Washington, DC: American Association of University

Women.

Hjortskov, N., Rissén, D., Blangsted, K., Fallentin, N., Lundberg, U., Søgaard, K. (2004). The

http://www.theesa.com/facts/pdfs/ESA_EF_2008.pdf

http://dx.doi.org/10.1016/0013-4694(94)00262-J

http://dx.doi.org/10.1016/0013-4694(94)00262-J

235


effect of mental stress on heart rate variability and blood pressure during computer work.

European Journal of Applied Physiology, 92(1‒2), 84‒89. doi: 10.1007/s00421-004-

1055-z

Houle, M. S., & Billman, G. E. (1999). Low-frequency component of the heart rate variability

spectrum: poor marker of sympathetic activity. American Journal of Physiology, 276(1),

H215‒223.

Hunicke, R., & Chapman, V. (2004). AI for dynamic difficulty adjustment in games. Challenges

in Game Artificial Intelligence AAAI, 91–96. doi: 10.1145/1178477.1178573

International Society for Presence Research. (2000). The Concept of Presence: Explication

Statement. Retrieved from http://ispr.info/

Jackson, S. A., & Eklund, R. C. (2002). Assessing Flow in Physical Activity: The Flow State

Scale-2 and Dispositional Flow Scale-2. Journal of Sport & Exercise Psychology, 24(2),

133‒150. doi: 10.1123/jsep.24.2.133

Järvelä, S., Ekman, I., Kivikangas, J. M., & Ravaja, N. (2015). Stimulus games. In P. Lankoski &

S. Björk (Eds.), Game Research Methods (pp. 193‒205). ETC Press. Retrieved from


Jennett, C., Cox, A. L., Cairns, P., Dhoparee, S., Epps, A., Tijs, T., & Walton, A. (2008).

Measuring and defining the experience of immersion in games. International Journal of

Human‒Computer Studies, 66(9), 641‒661. doi: 10.1016/j.ijhcs.2008.04.004

Jin, S. A. A. (2012). ‘Toward Integrative Models of Flow’: Effects of Performance, Skill,

Challenge, Playfulness, and Presence on Flow in Video Games. Broadcasting &

Electronic Media, 56(2), 169‒186. doi: 10.1080/08838151.2012.678516

Johnson, D., & Gardner, J. (2010). Personality, motivation and video games. Proceedings of the

22nd Conference of the Computer-Human Interaction Special Interest Group of Australia

on Computer-Human Interaction, 276‒279, Brisbane, Australia: ACM. doi:

10.1145/1952222.1952281

Johnson, D., Wyeth, P., Clark, M., & Watling, C. (2015). Cooperative game play with avatars and

agents: Differences in brain activity and the experience of play. Proceedings of the 33rd

Annual ACM Conference on Human Factors in Computing Systems, 3721‒3730, Seoul,

Republic of Korea. doi: 10.1145/2702123.2702468

Jönsson, P. (2007). Respiratory sinus arrhythmia as a function of state anxiety in healthy

individuals. International Journal of Psychophysiology, 63(1), 48‒54. doi:

10.1016/j.ijpsycho.2006.08.002

Keller, J., & Bless, H. (2008). Flow and regulatory compatibility: an experimental approach to the

flow model of intrinsic motivation. Personality & Social Psychology Bulletin, 34(2), 196–

209.

Keller, J., Bless, H., Blomann, F. & Kleinböhl, D. (2011). Physiological Aspects of Flow

Experiences: Skills-Demand Compatibility Effects on Heart Rate Variability and Salivary

Cortisol. Journal of Experimental Social Psychology, 47 (4), 849‒852. doi:

10.1016/j.jesp.2011.02.004

Keller, J., & Landhäußer, A. (2012). The flow model revisited. In S. Engeser (Ed.), Advances in

Flow Research (pp. 51–64). New York, NY: Springer.

http://doi.org/10.1145/1178477.1178573

http://dx.doi.org/10.1016/j.ijhcs.2008.04.004

http://dx.doi.org/10.1145/1952222.1952281

http://dx.doi.org/10.1145/1952222.1952281

http://dx.doi.org/10.1145/2702123.2702468

https://dx.doi.org/10.1016/j.ijpsycho.2006.08.002

236


Kirmizi-Alsan, E., Bayraktaroglu, Z., Gurvit, H., Keskin, Y. H., Emre, M., & Demiralp, T. (2006).

Comparative analysis of event-related potentials during Go/NoGo and CPT:

decomposition of eletrophysiological markers of response inhibition and sustained

attention. Brain Research, 1104(1), 114‒128. doi: 10.1016/j.brainres.2006.03.010

Kivikangas, J. M. (2006). Psychophysiology of Flow Experience: An Explorative Study (Master’s

thesis). Retrieved from

http://ethesis.helsinki.fi/julkaisut/kay/psyko/pg/kivikangas/psychoph.pdf

Kivikangas, J. M., Chanel, G., Cowley, B., Ekman, I., Salminen, M., Järvelä, S., & Ravaja, N.

(2011). A review of the use of psychophysiological methods in game research. Journal

of Gaming & Virtual Worlds, 3(3), 181–199. doi: 10.1386/jgvw.3.3.181_1

Kneer, J., Elson, M., & Knapp, F. (2016). Fight fire with rainbows: The effect of displayed

violence, difficulty, and performance in digital games on affect, aggression, and

physiological arousal. Computers in Human Behavior, 54(c), 142‒148. doi:

10.1016/j.chb.2015.07.034

Krygier, J. R., Heathers, J. A., Shahrestani, S., Abbott, M., Gross, J. J., & Kemp, A. H. (2013).

Mindfulness meditation, well-being, and heart rate variability: a preliminary investigation

into the impact of intensive Vipassana meditation. Intl. Journal of Psychophysiology,

89(3), 305‒313. doi: 10.1016/j.ijpsycho.2013.06.017

Lacey, J. I. (1967). Somatic response patterning and stress: Some revisions of activation theory.

In M. H. Appley & R. Trumbull (Eds.), Psychological Stress (pp. 14‒37). New York, NY:

Appleton Century Crofts.

Laerd Statistics (2015). One-way MANOVA using SPSS Statistics. Statistical tutorials and

software guides. Retrieved from https://statistics.laerd.com/

Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O. (1993). Looking at pictures:

affective, facial, visceral, and behavioral reactions. Psychophysiology, 30(3), 261–273.

doi: 10.1111/j.1469-8986.1993.tb03352.x

Lang, P.J. (1995). The Emotion Probe. Studies of motivation and attention. American

Psychologist, 50(5), 372‒385. doi: 10.1037/0003-066X.50.5.372

Lankoski, P. & Björk, S. (2015). Game Research Methods [ETC Press]. Retrieved from


Lehmann, D., Faber, P. L., Tei, S., Pascual-Marqui, R. D., Milz, P., & Kochi, K. (2012). Reduced

functional connectivity between cortical sources in five mediation traditions detected with

lagged coherence using EEG tomography. NeuroImage, 60(2), 1574‒1586. doi:

10.1016/j.neuroimage.2012.01.042

Lieberoth, A., & Roepstorff, A. (2015). Mixed methods in game research. In P. Lankoski & S.

Björk (Eds.), Game Research Methods (pp. 93‒116). ETC Press. Retrieved from


Lim, S., & Reeves, B. (2010). Computer agents versus avatars: Responses to interactive game

characters controlled by a computer or other player. International Journal of Human-

Computer Studies, 68(1-2), 57‒68. doi: 10.1016/j.ijhcs.2009.09.008

Lomas, D., Patel, K., Forlizzi, J. L., & Koedinger, K. R. (2013). Optimizing challenge in an

educational game using large-scale design experiences. CHI ’13 Proceedings, 89‒98,

Paris, France. doi: 10.1145/2470654.2470668

https://dx.doi.org/10.1016/j.brainres.2006.03.010

http://doi.org/10.1386/jgvw.3.3.181_1

http://dx.doi.org/10.1016/j.chb.2015.07.034



http://dx.doi.org/10.1016/j.ijhcs.2009.09.008

https://doi.org/10.1145/2470654.2470668

237


Lombard, M., & Ditton, T. (1997). At the heart of it all: The concept of presence. Journal of

Computer-Mediated Communication, 3(2). doi: 10.1111/j.1083-6101.1997.tb00072.x

Mandryk, R. L., & Inkpen, K. M. (2004). Physiological indicators for the evaluation of collocated

collaborative play. 2004 ACM conference on Computer Supported Cooperative Work,

102‒111, Chicago, Illinois, USA. doi: 10.1145/1031607.1031625

Mandryk, R. L., Atkins, M. S., & Inkpen, K. (2006a). A continuous and objective evaluation of

emotional experience with interactive play environments. SIGCHI Conference on Human

Factors in Computing Systems, Montréal, Québec, Canada, 1027‒1036. doi:

10.1145/1124772.1124926

Mandryk, R. L., Inkpen, K. M., & Calvert, T. W. (2006b). Using psychophysiological techniques

to measure user experience with entertainment technologies. Behaviour & Information

Technology, 25(2), 141‒158. doi: 10.1080/01449290500331156

Mandryk, R. L., & Atkins, M. S. (2007). A fuzzy physiological approach for continuously

modeling emotion during interaction with play technologies. International Journal of

Human Computer Studies, 65(4), 329–347. doi: 10.1016/j.ijhcs.2006.11.011

Mandryk, R. L. (2008). Physiological Measures for Game Evaluation. In M. Kaufmann, K.

Isbister, & N. Shaffer (Eds.), Game Usability: Advice from the Experts for Advancing the

Player Experience (pp. 207‒235). Burlington, MA: CRC Press.

Maskeliunas, R., Damasevicius, R., Martisius, I., & Vasiljevas, M. (2016). Consumer-grade EEG

devices: are they usable for control tasks? PeerJ, 4, e1746.

ttp://doi.org/10.7717/peerj.1746

Melillo, P., Bracale, M., & Pecchia, L. (2011). Nonlinear Heart Rate Variability features for real-

life stress detection. Case study: students under stress due to university examination.

BioMedical Engineering OnLine, 10(96). doi: doi: 10.1186/1475-925X-10‒96

Mirza-babaei, P., Nacke, L. E., Gregory, J., Collins, N., & Fitzpatrick, G. (2013). How Does It

Play Better ? Exploring User Testing and Biometric Storyboards in Games User Research.

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI

’13), 1499‒1508, Paris, France. doi: 10.1145/2470654.2466200

Mitchell, D. J., McNaughton, N., Flanagan, D., & Kirk, I. J. (2008). Frontal-midline theta from

the perspective of hippocampal ‘theta’. Progress in Neurobiology, 86(3), 156‒185. doi:

10.1016/j.pneurobio.2008.09.005

Nacke, L. & Lindley, C.A. (2008). Boredom, Immersion, Flow – a pilot study investigating player

experience. Measurement, 24(4), 1‒5. doi: 10.1073/pnas.1007983107

Nacke, L., & Lindley, C. A. (2008). Flow and immersion in first-person shooters: measuring the

player’s gameplay experience. In Proceedings of the 2008 Conference on Future Play:

Research, Play, Share, 81–88, Toronto, ON. doi: 10.1145/1496984.1496998

Nacke, L., Drachen, A., Kuikkaniemi, K., Niesenhaus, J., Korhonen, H. J., van den Hoogen, W.

M. … de Kort, Y. A. W. (2009). Playability and Player Experience Research. In

Proceedings of DiGRA 2009, 1‒5. ISBN / ISNN: ISSN 2342-9666

Nacke, L.E. (2010). Wiimote vs. controller: electroencephalographic measurement of affective

gameplay interaction. In Proceedings of the International Academic Conference on the

http://dx.doi.org/10.1145/1031607.1031625

http://dx.doi.org/10.1145/1124772.1124926

http://dx.doi.org/10.1145/1124772.1124926

http://doi.org/10.1016/j.ijhcs.2006.11.011

https://doi.org/10.7717/peerj.1746

238


Future of Game Design and Technology, 159‒166, Vancouver, BC. doi:

10.1145/1920778.1920801

Nacke, L.E., Grimshaw, M.N. & Lindley, C.A. (2010a). More than a feeling: Measurement of

sonic user experience and psychophysiology in a first-person shooter game. Interacting

with Computers, 22(5), 336‒343. doi: 10.1016/j.intcom.2010.04.005

Nacke, L., Drachen, A., & Göbel, S. (2010b). Methods for Evaluating Gameplay experience in a

Serious Gaming Context. International Journal of Computer Science in Sport, 9(2), 40‒

51. Retrieved from: http://hci.usask.ca/publications/view.php?id=174

Nacke, L., Kalyn, M., Lough, C., & Mandryk, R. L. (2011). Biofeedback game design: using direct

and indirect physiological control to enhance game interaction. Proceedings of the 2011

annual conference on Human factors in computing systems (CHI ’11), 103–112,

Vancouver, BC. doi: 10.1145/1978942.1978958

Nacke, L. E. (2013). An Introduction to Physiological Player Metrics for Evaluating Games. In

M. S. El-Nasr, A. Drachen, & A. Canossa (Eds.), Game Analytics: Maximizing the Value

of Player Data (pp. 585‒619). London, UK: Springer London.

Nakamura, J., & Csikszentmihalyi, M. (2002). The concept of flow. In C. R. Snyder & S. J. Lopez

(Eds.), Handbook of positive psychology (pp. 89–105). Oxford University Press.

Negini, F., Mandryk, R. L., & Stanley, K. G. (2014). Using affective states to adapt characters,

NPCs, and the environment in a first-person shooter game. In Games Media

Entertainment (GEM), 2014 IEEE, 1-8, Toronto, Canada. doi:

10.1109/GEM.2014.7048094

Newell, G. (2008, December 24). Gabe Newell Writes for Edge. Edge Online. Retrieved from

http://www.edge-online.com/opinion/gabe-newell-writes-edge

Nickel, P., & Nachreiner, F. (2003). Sensitivity and Diagnostics of the 0.1-Hz Component of Heart

Rate Variability as an Indicator of Mental Workload. Human Factors, 45(4), 575‒590.

doi: 10.1518/hfes.45.4.575.27094

Öhmann, A., Hamm, A., & Hugdahl, K. (2000). Cognition and the Automatic Nervous System:

Orienting, Anticipation, and Conditioning. In J. T. Cacioppo, L. G. Tassinary, & G. G.

Berntson (Eds.), Handbook of Psychophysiology (2nd ed.) (pp. 533‒575). New York, NY:

Cambridge University Press.

Patel, N. (2010, June 2.) Nintendo Wii Vitality Sensor. Engadget. Retrieved from

https://www.engadget.com/

Peifer, C. (2012). Psychophysiological Correlates of Flow-Experience. In S. Engeser (Ed.),

Advances in Flow Research (pp. 139‒164). New York, NY: Springer.

Phillips, C., Johnson, D., & Wyeth, P. (2013). Videogame reward types. Proceedings of the First

International Conference on Gameful Design, Research and Application (Gamification

’13), 103‒106, Toronto, ON. 10.1145/2583008.2583025

Preece, J., Rogers, Y. & Sharp, H. (2002). Interaction Design: Beyond Human‒Computer

Interaction. New York, NY: John Wiley & Sons, Inc.

Przybylski, A. K., Rigby, C. S., & Ryan, R. M. (2010). A motivational model of video game

engagement. Review of General Psychology, 14, 154–166. doi: 10.1037/a0019440

http://hci.usask.ca/publications/view.php?id=174

http://www.edge-online.com/opinion/gabe-newell-writes-edge

https://www.engadget.com/

http://doi.org/10.1037/a0019440

239


Przybylski, A. K., Weinstein, N., Murayama, K., Lynch, M. F., & Ryan, R. M. (2012). The ideal

self at play: The appeal of video games that let you be all you can be. Psychological

Science, 23 (1), 69‒76. doi: 10.1177/0956797611418676

Ravaja, N., Salminen, M., Holopainen, J., Saari, T., Laarni, J., & Järvinen, A. (2004). Emotional

response patterns and sense of presence during video games: Potential criterion variables

for game design. In NordiCHI ’04 Proceedings, Tampere, Finland, 339‒347. doi:

10.1145/1028014.1028068\

Ravaja, N., Saari, T., Salminen, M., Laarni, J., & Kallinen, K. (2006a). Phasic emotional reactions

to video game events: A psychophysiological investigation. Media Psychology, 8(4),

343‒367. doi: 10.1207/s1532785xmep0804_2

Ravaja, N., Saari, T., Turpeinen, M., Laarni, J., Salminen, M., & Kivikangas, M. (2006b). Spatial

presence and emotions during video game playing: Does it matter with whom you play?

Presence: Telepoerators & Virtual Environments, 15(4), 381‒392. doi: doi:

10.1162/pres.15.4.381

Ravaja, N., Turpeinen, M., Saari, T., Puttonen, S., & Keltikangas-Järvinen, L. (2008). The

psychophysiology of James Bond: Phasic emotional responses to violent video game

events. Emotion, 8(1), 114‒120. doi: 10.1037/1528-3542.8.1.114

Rheinberg, F., & Vollmeyer, R. (2003). Flow-Erleben in einem Computerspiel unter experimentell

variierten Bedingungen [Flow experience in a computer game under experimentally

controlled conditions]. Zeitschrift für Psychologie, 211, 161‒170. doi: 10.1026//0044-

3409.211.4.161

Rigby, S., & Ryan, R. (2007). The Player Experience of Needs Satisfaction (PENS): An applied

model and methodology for understanding key components of the player experience.

Retrieved from immersyve.com/PENS_Sept07.pdf

Russell, J.A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology,

39(6), 1161‒1178. doi: doi: 10.1037/h0077714

Russoniello, C. V., O'Brien, K., & Parks, J. M. (2009). EEG, HRV and psychological correlates

while playing Bejeweled II: A randomized controlled study. Annual Review of

Cybertherapy and Telemedicine, 7, 189‒192. http://www.ecu.edu/cshhp/

biofeedback/upload/EEG-HRV-and-Psychological-Correlates-while-Playing-

Bejeweled-II-A-Randomized-Controlled-Study.pdf

Ryan, R. M., & Deci, E. L. (2000). Self-Determination Theory and the Facilitation of Intrinsic

Motivation, Social Development, and Well-Being, 55(1), 68–78.

Ryan, R. M., Rigby, C., & Przybylski, A. (2006). The Motivational Pull of Video Games: A Self-

Determination Theory Approach. Motivation and Emotion, 30(4), 344–360. doi:

10.1007/s11031-006-9051-8

Saleem, M., Anderson, C. A., & Gentile, D. A. (2012). Effects of Prosocial, Neutral, and Violent

Video Games on College Students’ Affect. Aggressive Behavior, 38(4), 263‒271. doi:

10.1002/ab.21427

Salminen, M., & Ravaja, N. (2007). Oscillatory brain responses evoked by video game events:

The case of Super Monkey Ball 2. CyberPsychology & Behavior, 10(3), 330‒338. doi:

10.1089/cpb.2006.9947

Sanei, S. & Chambers, J. A. (2008). EEG Signal Processing. Wiley.

http://dx.doi.org/10.1177/0956797611418676

http://doi.org/10.1145/1028014.1028068

http://doi.org/10.1145/1028014.1028068

http://dx.doi.org/10.1207/s1532785xmep0804_2

http://dx.doi.org/10.1037/1528-3542.8.1.114

http://dx.doi.org/10.1026/0044-3409.211.4.161

http://dx.doi.org/10.1026/0044-3409.211.4.161

http://doi.org/10.1007/s11031-006-9051-8

http://doi.org/10.1007/s11031-006-9051-8

https://dx.doi.org/10.1002/ab.21427

240


Schier, M. A. (2000). Changes in EEG alpha power during simulated driving: a demonstration.

Intl. Journal of Psychophysiology, 37(2), 155‒162. doi: 10.1016/s0167-8760(00)00079-

9

Schubert, C., Lambertz, M., Nelesen, R. A., Bardwell, W., Choi, J. B., & Dimsdale, J. E. (2009).

Effects of stress on heart rate complexity – a comparison between short-term and long-

term chronic stress. Biological Psychology, 80(3), 325‒332. doi:

10.1016/j.biopsycho.2008.11.005

Sherry, J. L. (2004). Flow and Media Enjoyment. Communication Theory, 14(4), 328‒347. doi:

10.1111/j.1468-2885.2004.tb00318.x

Sherry, J. L., Greenberg, B. S., Lucas, K., & Lachlan, K. A. (2006). Video game uses and

gratifications as predictors of use and game preference. In P. Vorderer & J. Bryant (Eds.),

Playing Video Games: Motives, Responses, and Consequences (pp. 213‒224). New York,

NY: Routledge.

Sims, A. (1988). Symptoms in the Mind: An Introduction to Descriptive Psychopathology.

Saunders Ltd.

Skalski, P., Francis, D., Kushin, M., & Liu, Y. I. (2012). Need for Presence and Other Motivations

for Video Game Play across Genres. Proceedings of the ISPR ’12. Retrieved from

http://socialsciences.people.hawaii.edu/publications_lib/Dalisay.Need%20for%20Prese

nce.pdf

Stern, R. M, Ray, W. J, & Quigley, K. S. (2001). Psychophysiological Recording (2nd ed.). New

York, NY: Oxford University Press.

Sweetser, P., & Wyeth, P. (2005). GameFlow: a model for evaluating player enjoyment in games.

Computer Entertainment, 3(3), 3–24. doi: 10.1145/1077246.1077253

Szabo, A., & Gauvin, L. (1992). Mathematical performance before, during, and following cycling

at workloads of low and moderate intensity. Perceptual and Motor Skills, 75(3), 915‒918.

doi: 10.2466/pms.1992.75.3.915

Tabachnick, B. G., & Fidell, L. S. (2007). Using Multivariate Statistics (5th ed). Needham Heights,

MA: Allyn & Bacon, Inc.

Tarvainen, M. P., Niskanen, J-.P., Lipponen, J. A., Ranta-aho, P. O., Karjalainen, P. A. (2009).

Kubios HRV – A Software for Advanced Heart Rate Variability Analysis. Proceedings

of the 4th European Conference of the International Federation for Medical and

Biological Engineering, 1022‒1025, Antwerp, Belgium. doi: 10.1007/978-3-540-89208-

3_243

Tarvainen, M. P., Niskanen, J-.P., Lipponen, J. A., Ranta-aho, P. O., Karjalainen, P. A. (2014).

Kubios HRV – Heart rate variability analysis software. Computer Methods and Programs

in Biomedicine, 113(1), 210‒220. doi: 10.1016/j.cmpb.2013.07.024

Task Force of the European Society of Cardiology and the North American Society of Pacing and

Electrophysiology. (1996). Heart rate variability: Standards of measurement,

physiological interpretation, and clinical use. European Heart Journal, 17(3), 354‒381.

doi: https://doi.org/10.1161/01.CIR.93.5.1043

Tassinary, G. G., & Cacioppo, L. G. (2000). The Skeletomotor System: Surface

Electromyography. In J. T. Cacioppo, L. G. Tassinary, & G. G. Berntson (Eds.),

http://doi.org/10.1111/j.1468-2885.2004.tb00318.x

http://doi.org/10.1111/j.1468-2885.2004.tb00318.x

http://socialsciences.people.hawaii.edu/publications_lib/Dalisay.Need%20for%20Presence.pdf

http://socialsciences.people.hawaii.edu/publications_lib/Dalisay.Need%20for%20Presence.pdf

http://doi.org/10.1145/1077246.1077253

https://dx.doi.org/10.2466/pms.1992.75.3.915

241


Handbook of Psychophysiology (2nd ed.) (pp. 163‒199). New York, NY: Cambridge

University Press.

Tychsen, A., & Canossa, A. (2008). Defining personas in games using metrics. Proceedings of

Future Play ’08, 73‒80, Toronto, ON. doi: 10.1145/1496984.1496997

Valve Corporation. (2009). Left 4 Dead 2 [PC and XBox game].

van der Vijgh, B., Beun, R. J., Van Rood, M., & Werkhoven, P. (2015). Meta-analysis of digital

game and study characteristics eliciting physiological stress responses.

Psychophysiology, 52(8), 1080‒1098. doi: 10.1111/psyp.12431

Vasey, M. W., & Thayer, J. F. (1987). The Continuing Problem of False Positives in Repeated

Measures ANOVA in Psychophysiology: A Multivariate Solution. Psychophysiology,

24(4), 479‒486. doi: 10.1111/j.1469-8986.1987.tb00324.x

Vella, K., Johnson, D., & Hides, L. (2015). Indicators of wellbeing in recreational video game

players. Proceedings of the Annual Meeting of the Australian Special Interest Group for

Computer Human Interaction (OzCHI ’15), 613‒617, Melbourne, Australia. doi:

10.1145/2838739.2838818

Vella, K. (2016). The social context of video game play: Relationships with the player experience

and wellbeing (Doctoral thesis). Retrieved from https://eprints.qut.edu.au/95981/

Ward, R. D., & Marsden, P. H. (2003). Physiological responses to different WEB page designs.

International Journal of Human‒Computer Studies, 59(1‒2), 199‒212. doi:

10.1016/S1071-5819(03)00019-3

Wastell, D. G., & Newman, M. (1996). Stress, control and computer design: a psychophysiological

field study. Behaviour & Information Technology, 15(3), 183‒192. doi:

10.1080/014492996120247

Weber, R., Behr, K.-M., Tamborini, R., Ritterfield, U., & Mathiak, K. (2009). What do we really

know about first-person shooter games? An event-related, high-resolution content

analysis. Computer-Mediated Communication, 14(4), 1016‒1037. doi: 10.1111/j.1083-

6101.2009.01479.x

Weinreich, A., Strobach, T., & Schubert, T. (2014). Expertise in video game playing is associated

with reduced valence-concordant emotional expressivity. Psychophysiology, 52(1), 59‒

66. doi: 10.1111/psyp.12298

Whitham, E. M., Pope, K. J., Fitzgibbon, S. P., Lewis, T., Clark, C. R., Loveless, S., …

Willoughby, J. O. (2007). Scalp electrical recording during paralysis: Quantitative

evidence that EEG frequencies above 20 Hz are contaminated by EMG. Clinical

Neurophysiology, 118(8), 1877‒1888. doi: 10.1016/j.clinph.2007.04.027

Whitham, E. M., Lewis, T., Pope, K. J., Fitzgibbon, S. P., Clark, C. R., Loveless, S., …

Willoughby, J. O. (2008). Thinking activates EMG in scalp electrical recordings. Clinical

Neurophysiology, 119(5), 1166‒1175. doi: 10.1016/j.clinph.2008.01.024

Wiemeyer, J., Nacke, L., Moser, C., & Mueller, F. (2016). Player Experience. In R. Dörner, S.

Göbel, W. Effelsberg, & J. Wiemeyer (Eds.), Serious Games: Foundations, Concepts and

Practice (pp. 243–271). Springer International Publishing.

Wilson, G. M., & Sasse, M. A. (2000). Investigating the Impact of Audio Degradations on Users:

Subjective vs. Objective Assessment Methods. OZCHI 2000: Interfacing Reality in the

https://eprints.qut.edu.au/95981/

242


New Millenium, Sydney, Australia, 135‒142. Retrieved from:

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.15.8278&rep=rep1&type=pdf

Wirth, W., Hartmann, T., Böcking, S., Vorderer, P., Klimmt, C., Schramm, H., Saari, T., Laarni,

J., Ravaja, N., Gouveia, F. R., Biocca, F., Sacau, A., Jäncke, L., Baumgartner, T., &

Jäncke, P. (2007). A process model of the formation of spatial presence experiences.

Media Psychology, 9(3), 493‒525. doi: 10.1080/15213260701283079

Witvliet, C. V., & Vrana, S. R. (1995). Psyhcophysiological responses as indices of affect

dimensions. Psychophysiology, 32(5), 436‒443. doi: 10.1111/j.1469-

8986.1995.tb02094.x

Wu, S., & Lin, T. (2011). Exploring the use of physiology in adaptive game design. 2011

International Conference on Consumer Electronics, Communications and Networks

(CECNet), 1280‒1283, Xianning, China. doi: 10.1109/CECNET.2011.5768186

http://dx.doi.org/10.1080/15213260701283079

243


8 APPENDICES

8.1 APPENDIX A—FSS SAMPLE QUESTIONS

As the FSS and S FSS-2 are commercial scales, they cannot be published. Five

sample questions from the FSS are thus provided below. Items are scored using a 5-point

Likert scale from ‘strongly disagree’ to ‘strongly agree’.

I felt I was competent enough to meet the demands of the situation.

I did things spontaneously and automatically without having to think.

I had a strong sense of what I wanted to do.

I had a good idea about how well I was doing while I was involved in the

task/activity.

I was completely focused on the task at hand.

244


8.2 APPENDIX B—EXAMPLE PENS ITEMS

As the PENS is a commercial scale, it cannot be published. Three sample items

from the PENS are thus provided below. Items are scored using a 7-point Likert scale

from ‘do not agree’ to ‘strongly agree’.

Thinking about playing, reflect on your play experiences and rate your

agreement with the following statements:

I experienced a lot of freedom in the game (autonomy).

I feel very capable and effective when playing (competence).

When moving through the game world I feel as if I am actually there (presence).

245


8.3 APPENDIX C—IMI INTEREST/ENJOYMENT SUBSCALE ITEMS

246


8.4 APPENDIX D—DEMOGRAPHICS QUESTIONNAIRE

247


8.5 APPENDIX E—STUDY 2 SCRIPT

EXPERIMENT OVERVIEW

00.00 – 05:00: Introduction; brief; receive signed and dated consent forms

05.00 ‒ 10:00: Demographics questionnaire

10.00 – 12.00: Participant hands wash

12.00 – 50.00: Biometrics set-up (refer to set-up guide, below)

50.00 – 52.00: Baseline (FLAG: F1 START, F2 END)

52.00 – 57.00: Tutorial (FLAG: F7 START, F8 END)


59.00 – 70.00: First play session (FLAG: F5 START, F6 END)

70.00 – 77.00: First surveys


79.00 – 90.00: Second play session (FLAG: F5 START, F6 END)

90.00 – 97.00: Second surveys


99.00 – 110.00: Third play session (FLAG: F5 START, F6 END)

110.00 – 117.00: Third surveys


119.00 – 124.00: Remove electrodes, etc.

124.00 – 126.00: Debrief

PRE-EXPERIMENT

- Set up nurse’s trolley:

o EMG/ECG electrodes

o nuprep

o tape

o conductive gel

o scissors

o alcohol wipes

o cotton buds

o gauze pads.

- Soak all EEG electrode pads.

- Ensure sink is filled and running.

- Make sure Left 4 Dead 2 doesn’t need to patch (patch if required).

- Initiate Sequencer and pre-fill in participant ID.

- Write participant number on post-it note; attach to participant desk.

- Have consent form + pen ready.

- Launch AcqKnowledge

248


o Recent > AllChannels.gtl.

PARTICIPANT ARRIVAL:

- Welcome them and read intro script.

- Receive signed and dated consent forms.

- Get them to answer demographics questionnaire.

- Follow set-up below. Tell the participant what you are doing at all times (e.g., which

area you are abrading next). Keep them engaged and comfortable.

Biometric Set-up Guide

EDA:

1. Get participants to wash hands; ensure they are dried thoroughly.

2. Attach disposable EL507 electrodes to the palm (green collared electrode on the

hypothenar—or pinkie side of the palm).

3. Secure electrodes with tape.

EMG1 (CS):

1. Locate the corrugator supercilii (refer to figure). You may need to touch the

participant’s brow area and get them to emote (frown, raise eyebrows) to find it.

2. Abrade area (nuprep, cotton bud).

3. Wipe area clean (alcohol swipe, clean gauze pad). Note: ask participant to close eyes for

alcohol swipe.

4. Take an ADD204 collar and peel the bottom waxed paper strip and apply carefully to an

EL254S Ag-AgCl electrode.

5. Fill the cavity with electrode gel (GEL100), avoiding air bubbles.

6. Level off excess electrode gel (not with fingers).

7. Repeat for a second EL254S.

8. Repeat for a ground (black lead) EL254.

9. Attach ground electrode to centre of forehead.

10. Attach the pair of EL254S to the CS location.

11. Lead electrode wires behind participant ear to prevent vision obfuscation.

12. Secure with tape.

13. Clip on EMG1 extender to the back of participant’s collar.

14. Plug ground electrode into GND.

15. For electrode closest to nose, plug white lead into VIN+, black lead into corresponding

shield port.

16. Repeat for second EL254S electrode, with VINÈ.

17. Check AcqKnowledge for signal.

249


18. Check Checktrode (> 10 kiloohms)

EMG2 (OO):

1. Locate the orbicularis oculi (refer to figure). You may need to touch the participant’s

eye area and get them to emote (smile, laugh) to find it.

2. Steps 2‒7 in EMG CS.

3. Attach the pair of EL254S to the OO location.

4. Lead electrode wires behind participant ear to prevent vision obfuscation.


6. Clip on EMG2 extender to the back of participant’s collar.

7. Steps 15–18 in EMG CS.

ECG:

1. Abrade skin approximately 3–5 cm below right collarbone (refer to figure).

2. Abrade skin approximately 3 cm above participant’s lowest left-side rib (aligning with

their elbow—refer to figure).

3. Wipe both areas clean with alcohol wipe and clean gauze pad. Note: ask participant to

close eyes for alcohol swipe.

4. Steps 4–7 of EMG CS.

5. Attach one EL254S electrode each to prepared sites.


7. Clip the ECG extender onto the back of the participant’s collar.

8. For collarbone electrode, white lead into VIN+ and black lead into SHIELD.

9. For ribcage electrode, white lead into VIN‒ and black lead into shield.

10. Check AcqKnowledge for signal.

11. Check Checktrode (> 10 kiloohms).

EEG:

1. Launch TestBench software.

2. Place EEG on head; ensure electrodes are in correct position (refer to figure).

3. Adjust headset until all electrodes make contact (green signal on TestBench).

4. If required, re-soak electrode pads; may need to soak participant hair (note: ask

permission first, ask them to shut their eyes).

5. Repeat until all green signals acquired.

COMFORT CHECK: Once all instruments are set up, ensure they are comfortable (no pressure,

poking or sensation of electrode detaching).

DATA COLLECTION + SCRIPTS

250


- Start screen recording.

- Initiate Sequencer (press ‘Next’).

DURING BASELINES:

First time: ‘The screen is about to go blank for a short period of time. When it does, please look

at the screen and try to relax. While doing so, try to remain relatively still and not tap your

fingers, feet and so on.’

All others: ‘The screen is about to go blank again. Again, I ask that you try to relax and not tap

fingers, feet and so on.’

DURING TUTORIAL:

‘The objective of the game is to fill a generator with fuel canisters. Enough fuel will lower a

bridge and grant you access to a car, completing the mission. The fuel canisters are scattered

throughout the map. While exploring, you will likely encounter enemy zombies that you may

need to shoot.

I ask that you please do not change any settings such as key bindings.’

BEFORE FIRST PLAY SESSION:

‘I ask that you try and complete the mission to the best of your ability. If you die at any point

during any of the play sessions, just keep playing—the level will restart.

Above all else, your main goal here is to try to have fun.’

POST-EXPERIMENT

- Dispose of all rubbish.

- Clean EMG/ECG electrodes with toothbrush + tepid water.

- Disinfect EMG/ECG electrodes for 5 minutes in disinfectant.

- Rinse EMG/ECG electrodes and hang to dry.

- Move data onto thumb drive.

- Put EEG on charge.

251


8.6 APPENDIX F—ELECTROENCEPHALOGRAPHY EEG OUTLIERS

All outliers for untransformed EEG data. In order: AF4 Alpha, Beta, Theta; O2 Alpha,

Beta, Theta.

252


253


254


8.7 APPENDIX G—BOREDOM, BALANCE, AND OVERLOAD PLAY

CONDITION VIDEOS Links to video footage demonstrating play from the Boredom, Balance, and Overload

play conditions are contained below.

Boredom: https://www.youtube.com/watch?v=Rr1vr7Drguk

Balance: https://www.youtube.com/watch?v=s6dwu0zs8No

Overload: https://www.youtube.com/watch?v=rYqRh4zXxwc

https://www.youtube.com/watch?v=Rr1vr7Drguk

https://www.youtube.com/watch?v=s6dwu0zs8No

https://www.youtube.com/watch?v=rYqRh4zXxwc

Documents

THE PSYCHOPHYSIOLOGICAL - QUT · The psychophysiological method presents one approach for evaluating this player experience. A variety of psychophysiological investigations of the