Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Emotion control of player characters: creating an emotionally responsive game
(USC Thesis Other)
Emotion control of player characters: creating an emotionally responsive game
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
EMOTION CONTROL OF PLAYER CHARACTERS:
CREATING AN EMOTIONALLY RESPONSIVE GAME
by
Brandi Wilcox
A Thesis Presented to the
FACULTY OF THE USC SCHOOL OF CINEMATIC ARTS
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF FINE ARTS
(INTERACTIVE MEDIA)
May 2010
Copyright 2010 Brandi Wilcox
Dedication
This thesis is dedicated to Dan and Katrina Wilcox for all their help, support, and cookies. An
über-fantabulous thanks to Dan for all his insight into this project and encouragement. This thesis
is also dedicated to Shawna Wilcox, thanks for being a friendly ear!
ii
Acknowledgements
A hearty and well-felt thanks to the people who advised on this project and gave much needed
insightful feedback, namely Mike Ellis, Joel Goodsell, Paul De Meo, and Perry Hoberman. This
project would not have happened without you. Thanks goes out to Anne Balsamo, especially for
her help with this paper, Mark Bolas, and Steve Anderson for all their help in the development of
this project. Additional thanks to Tracy Fullerton, for teaching what game design actually is. A big
thanks to Sam Pierce for all his hard work and the fantastic music. A special thank you to all the
play-testers, all of which gave valuable feedback into making the project a better experience. And
lastly, thank you to the graduate students of 2010: Lulu Cao, Alaʼ Diab, Bryan Jaycox, Cynthia
Nie, Taiyoung Ryu, Nahil Sharkasi, and Peter Van Dyke. You are all awesome and it has been a
great learning experience and a blast spending the last 3 years together!
iii
Table of Contents
Dedication ......................................................................................................................." ii
Acknowledgements ........................................................................................................." iii
List of Figures ................................................................................................................." v
Abbreviations .................................................................................................................." vi
Abstract ..........................................................................................................................." vii
Introduction ....................................................................................................................." 1
Project Description .........................................................................................................." 3
Emotion Model ................................................................................................................" 5
Prior Art - Games ............................................................................................................" 10
Prior Art - Papers ............................................................................................................" 22
User Testing & Evaluation ..............................................................................................." 25
Discussion ......................................................................................................................." 37
Conclusion ......................................................................................................................." 39
Sorted Bibliography .........................................................................................................." 40
Comprehensive Bibliography ..........................................................................................." 44
iv
List of Figures
Figure 1: Screenshot of Crimson Night ...........................................................................! 4
Figure 2: Plutchikʼs emotion wheel (Plutchik 2001, 349) ................................................." 7
Figure 3: Screenshot of the first prototype, which was purely
" conversational with only mood control for the mechanics .................................." 27
Figure 4: Screenshot of a conversation in Crimson Night ...............................................! 28
Figure 5: Close-up of the “action-wheel” used in an early prototype .............................." 29
Figure 6: A screenshot of an early prototype identifying the “mood-wheel” ...................." 31
Figure 7: A table of the moods displayed by the face-icon to denote
" the PCʼs current mood in the “mood-wheel” used
" in an early prototype .........................................................................................." 32
Figure 8: Mood screen effects ........................................................................................" 34
v
Abbreviations
Standard Abbreviations in games:
" IF = Interactive Fiction
! NPC = Non-Playable Character
" RPG = Role-Playing Game
" PC = Player Character
Standard Abbreviations for platforms (only used in the Bibliography):
" PC = Personal Computer, with a Windows operating system
" MAC = an Apple Macintosh with an Apple operating system
" PS1 = the original Sony Playstation
" PS2 = Sony Playstation 2
" PS3 = Sony Playstation 3
" SNES = Super Nintendo Entertainment System
" Xbox = Microsoft Xbox
" Xbox 360 = Microsoft Xbox 360
"
vi
Abstract
The gaming industry is diligently working to find ways to emotionally immerse players in games.
Their goal is to evoke emotions in the player, such as empathy, to make the player feel more
genuinely involved in the game experience. A large body of work, both theoretical and physical,
has been done in this regard. However, in order to have an emotionally immersive game, the
player and the game must be involved in an emotion dialogue. It must both evoke emotions
(emotion output) and receive emotions (emotion input). To date, emotion input has been largely
ignored by game designers, leaving games that “talk” emotionally at the player but do not “listen”
to the player. This thesis explores emotion input by designing a game that prompts for and
responds to player input of emotions to create a game that listens to the player emotionally. The
hypothesis is that the capability to input emotions will create an emotionally responsive world that
will facilitate player involvement and immersion.
vii
Introduction
Games no longer have to be a simple experience, a linear track where the player has no effect on
the game experience. There has been an explosion of games that allow the player to make
choices and experience the consequences of these choices. Some games, like Mass Effect
(BioWare 2007), even have these repercussions carry through to their sequels creating extremely
long-term consequences. No matter how far reaching the consequences, most of these “choice-
based” games merely present the player with action choices that allow the player to select from
only a few alternatives.
Allowing the player to supply emotion input to the game and have the game be affected by this
information is a reasonable addition to the design of choice-driven games. Adding another layer
of player input and game feedback can further enrich the game experience and make for a more
responsive and believable world. It allows players to create a more realistic “puppet” of the
Player Character (PC). Previously, the player only had control of the external facets of the PC:
their position, movement, actions, physical appearance and choices. Emotions add some of the
internal components, allowing for a more fully realized PC and facilitating player role-playing.
With the existence of biosensors and brain-computer interfaces, it is possible to attempt to read
the playerʼs emotional state directly. However, it is imprecise; is a racing heart a sign of fear or
excitement? Is an increase in sweat a sign of being nervous or a result of an increase in ambient
temperature? Trying to understand a playerʼs real emotional state is a difficult prospect and
outside of the scope of this project. The goal is to make a game that listens to the emotion data
from the player and is not concerned with surveying all the methods to obtain this data.
Additionally, it is unnecessary to use the playerʼs real emotions to use emotion input and to
1
explore its effectiveness. The goal is to have the player use emotion input to explore a story
space and as a tool for role-playing.
Current games are not totally devoid of PC emotions, but they typically prohibit player control
over these emotions. Cloud, the PC from Final Fantasy VII (Square 1997), clearly becomes
emotionally upset with the death of Aeris, but this is just part of the story. No matter the playerʼs
feelings on the matter, Cloud will always mourn the loss of the healer. Giving the player the ability
to manipulate their PCʼs emotions will allow them to shape the story and game experience in new
and varied ways, creating a much more layered, responsive, and immersive experience. In the
previous example, giving the player the ability to manipulate Cloudʼs emotions could have
resulted in Cloud being angry over her death, ambivalent, or even happy, severely impacting the
story. This interaction can add depth and story-customization to any story-driven game, and
when perfected could even be part of the solution for emotionally immersive games.
The purpose behind this thesis and its accompanying interactive project, Crimson Night, is to
explore methods that 1) allow player manipulation of the PCʼs emotions and 2) provides a way for
the game to respond to that input. By making a game that listens for emotion input from the
player, the game and its story becomes more malleable by the player and thus leads to an
interesting and enjoyable experience that reacts to emotion input.
2
Project Description
Crimson Night, the accompanying interactive project, is a story-driven adventure game developed
to test and refine the methods of emotion input to create an emotionally responsive game. It is a
game where the PC is caught in an emotional dilemma about who she is, who to trust, and where
her loyalties lie. Her emotional state transforms how she views her world and how she interacts
with NPCs as well as how they react to her. It was consciously tailored to have no “correct”
responses or choices. Ultimately, the story culminates in the revelation of her past, tailored to the
choices the player made for the PCʼs actions and moods.
Player interactions are with the mouse and keyboard or a Microsoft USB Xbox Controller. The
game was developed with Adobe Flash and as such can be played on a Windows, Macintosh, or
LINUX computer. Using Flash as the development environment also allowed for rapid prototypes
and iterative changes as the game was developed.
The story revolves around the PC, Alyn Britemore, a soldier for the Unified Territories (the UT),
who is captured by the enemy Doman nation. In custody, she struggles with what she knows
about herself and what her captors present to her. Alyn had previously suffered a near-fatal
wound that has given her some memory damage, but the Domani claim that the real Alyn is dead
and they just moved her memory into the Domani Reyvka Reshon. Reyvkaʼs parents try to
convince the orphaned Alyn that she is their daughter while the similarly captured Juri, Alynʼs
beloved fiancé, tries to keep her grounded to her remembered past. Is the PC Alyn or Reyvka?
Is the PC in Doman or still in the UT? Who is trying to save her and who is trying to turn her to
their advantage? Player choices of emotions and actions dictate which one of these are “true”
and determines the consequences.
3
Player interactions with the game are purely explorative and conversational, with a focus on
player manipulations of the PCʼs emotional state. There are no puzzles or combat in this small
sample of the game as they are unnecessary and distracting to the goal of exploring player
control of PC emotions and the gameʼs responses. The PCʼs emotional state affects other NPCʼs
emotions, NPC dialogue, and NPC opinions of and responses to the PC. It also affects what
actions the PC can perform, PC opinions on objects and people, what PC dialogue options are
available, and influences the ending.
Figure 1: Screenshot of Crimson Night
4
Emotion Model
A simple emotional model was needed to be the base for the gameʼs emotion input mechanic.
The study of emotions is a deep and complex field as the origins and purpose of emotions is
debatable. Theories of emotion are categorized into one of three types: evolutionary (emotions
are a result of natural selection), social (emotions are social constructions, learned through social
interactions), and internal (emotions caused from biological processes) (Johnson 2009).
However, the reasons behind emotions are unimportant for the gameʼs model. All that was
needed was a small group of “base” emotions so that the game did not have to accept too many
emotions, and that these emotions were understandable and clear to a player. There is no
definitive taxonomy of emotions, so there are different “base” or “primary” emotions depending on
the view or psychologist proposing them.
René Descartes wrote in The Passions of the Soul about there being “six primitive passions” of
admiration, love, hatred, desire, joy, and sadness, and that all other feelings are combinations of
these (Descartes 1650, 54). These “passions” are analogous to the modern concept of emotions,
but Descartes described them as emotions of the soul that were caused by “spirits” in the body
(Descartes 1650, 23). To “love” or “hate” someone or something was too simplistic as it can be
reduced to a “agree with” or “disagree with” system when referring to conversation topics.
“Admiration” would also be difficult to utilize in the game, as it is a fairly subtle emotion. “Desire”
means to want something, which was deemed not useful for the purposes of this project as
controlling what the PC wants was deemed more of an aspiration or needs system rather than an
emotion system. Therefore, Descartesʼ classification of basic emotions was rejected.
5
In 1677, Baruch Spinoza argued against Descartes theory of emotions in Ethics (Spinoza 1997,
Part III, Preface). In Ethics, Spinoza gives pleasure, pain, and desire as the three primary
emotions (Spinoza 1997, Part III, Definitions of the Emotions, IV). Pleasure and pain are a
response to generally physical stimulus however, like a hug or falling down. It does not lend itself
well as an emotion system for interacting with NPCs, especially through conversation. A topic
may be “painful” or “pleasurable” to talk about, but how would this be expressed and how would
the NPCs understand this? And “desire” would still be a problem for emulation as in the
Descartes model. It is also missing “anger” which is one emotion that would be beneficial to have
in the game, as it is simple and well-recognized by players. Therefore, Spinozaʼs set of base
emotions was unacceptable.
There is the minimalist view of there being only three core emotions: terror, ecstasy, and despair
(Jensen 2000). All other emotions are derived by a different degree of one of the core emotions
or from a combination of core emotions. This was unsatisfactory as it was not comprehensive
enough; it boiled down to the player picking from being happy, afraid, or sad. Anger is a common
emotion and one players will want to be able to use, but to use this model and have anger meant
doing combinations of the base emotions which would exponentially increase the number of
emotions used.
Paul Ekman ran an experiment of showing images of facial expressions of varying emotions to
seven different cultures including two isolated cultures in New Guinea who had little-to-no
exposure to the rest of the world (Ekman 1972 , 260-276). Ekman found some emotional facial
expressions were universal as the different cultures would positively identify them in photographs.
He concluded that the emotions of these universal facial expressions were the basic emotions of
happiness, sadness, anger, fear, disgust, and surprise (Ekman 1972, 278). This would be a good
6
model, provided “disgust” and “surprise” were dropped from the model as they are too difficult to
simulate and express in the game, which has a relatively small scope.
Robert Plutchik, an evolutionary emotion theorist, created a list of primary emotions based on
animal adaptations that he theorized occurred 600 million years ago (Johnson 2009). From a list
of core adaptive behaviors, he created a list of primary emotions associated with those behaviors,
which were anger, fear, joy, sadness, acceptance, disgust, anticipation, and surprise (Johnson
2009). All emotions are either varying degrees of these primary eight or combinations of them
(Plutchik 2001, 350). Further, he postulated that these eight primary emotions can be
conceptualized as being four pairs of polar-opposite emotions: fear versus anger, sadness versus
joy, anticipation versus surprise, acceptance versus disgust (Plutchik 2001, 349). He also
created an emotion wheel (Figure 2) to graphically display his emotion model. “Disgust,”
Figure 2: Plutchikʼs emotion wheel (Plutchik 2001, 349)
7
“acceptance,” “anticipation,” and “surprise” are too subtle and difficult to express in a game of this
caliber. Because the combination of emotions was not to be emulated, it was acceptable to drop
these emotions from the model with no ill effect. Plutchikʼs model for the base emotions is similar
to Ekmanʼs model except for the addition of the “acceptance” and “anticipation” emotions as well
as the concept of the emotions being arranged in opposite-pairs. However, it is this arrangement
of emotions into opposite-pairs that makes this model the ideal one for the emotion mechanic in
the game.
As a game mechanic, being able to reduce the emotional spectrum into two axes make the
emotion mechanic not only simpler from a design point of view but also from the playerʼs point of
view. Also, having each primary emotion have several different degrees of intensity would lend
itself to making emotion changes more subtle and realistic while also helping to keep the PC from
making extreme mood swings. The names of these emotions at different degrees (e.g.
“apprehension” and “terror” for fear) were not used, instead the prefixes of “mild,” “moderate,” and
“extreme” were used (e.g., “mild fear,” “moderate fear,” and “extreme fear”). This would make the
degrees clear as well as maintaining reference to the base emotion.
Another change to Plutchikʼs model was due to the fact it was missing a “neutral” or “emotionless”
state, which is understandable as the lack of emotion can hardly be an emotion itself. From a
game design point of view, it was necessary to have an emotion option that meant lacking any
perceivable emotion. This state is there for players to use when they do not want the PC to have
any particular emotional response or if the player does not want to use any of the available
emotions. It would also be unfair to start the PC in any specific emotion, so the neutral state
serves as the starting emotion position as well. The neutral state also serves as a buffer when
going from one emotion to its opposite, such that an angry PC must become neutral before
becoming afraid.
8
To keep the amount of acceptable emotions manageable, especially for a game of this scope, a
limit was placed on the variety of emotions available. Only a sub-set of primary emotions would
be used and the full emotional spectrum that comes from combinations of these primary emotions
would be ignored. For this purpose, the emotion model used in the game is based on Plutchikʼs
primary emotions, except for the removal of the “anticipation-surprise” and “disgust-acceptance”
opposite pairs and the addition of a “neutral” emotion. As in Plutchikʼs model, anger and fear are
treated as opposites, as are joy and sadness. Each of the non-neutral emotions were divided into
three degrees of “mild,” “moderate,” and “extreme” to make a total of 13 possible PC emotions.
This gives a good range of emotions and emotion intensities without becoming too overwhelming
for the game designer or the player. These 13 emotions are a sufficient set of emotions to
explore emotion input with and show the viability of this method to engage the player.
9
Prior Art - Games
Games that deal with emotions fall into two main categories: games with emotion input and
games with emotion output. Games with emotion input are those that allow the player control
over in-game emotions, and games with emotion output are those that elicit emotions in the
player. There are many games that involve emotion and many ways that they accomplish these
emotions. However, there are few that allow for player manipulation of the PCʼs emotions and
none of those to the degree of subtlety and consequences that are necessary to create the
emotionally responsive game this thesis attempts to explore.
Games with Emotion Input
Games with emotion input allow the player the ability to directly or indirectly affect emotions of
characters in the game. There are only two types of characters in a game, playable characters
(PCs), and non-playable characters (NPCs), so this category can be broken down into two
subcategories: games that affect PC emotions and games that affect NPC emotions.
Affecting PC Emotions
This thesis specifically explores this subcategory of games that allow players the ability to affect
PC emotions. There are few games published that have any sort of emotion control of the player
character, and these are mostly to a limited degree. The Act (Cecropia 2007) is one such coin-op
game. In The Act, the player uses a single rotating knob to manipulate a particular behavior or
emotion, such as sense of humor or courage, of the PC Edgar, with the goal to maintain the right
degree to pass the level. This is an uninteresting emotional control because it is a means to an
end; player control results in only failing or passing a level. This control adds nothing but novelty
10
to the player experience and therefore is not a good example of having games with player
manipulation of the PCʼs emotions, which requires nontrivial emotional choice.
Super Princess Peach (Nintendo 2006) has a mechanic where the player can make the PC,
Princess Peach, take on different emotions to have different abilities for traversal, solving
puzzles, and defeating enemies. These emotions are Joy, Rage, Calm, and Gloom. Joy gives
her the ability of wind, Rage gives her the ability of fire, Calm gives her a healing ability, Gloom
gives her the ability of water through her tears. A “Vibe” gauge depletes as players use emotions,
and must be refilled by collecting special crystals. This is not an emotion mechanic that is used
for story purposes, but rather as tools for puzzle solving. Overly dramatic depictions of these
emotions -- in Gloom, Princess Peach becomes a fountain of tears -- and quick emotion changes
leads to Princess Peach to appear emotionally unstable, which many players take as a sexist
stereotype about women and their emotions. As a means to an end, the emotion input content in
Super Princess Peach is not interesting and was detrimental in the characterization of Princess
Peach.
In both Blade Runner (Westwood Studios 1997) and The X-Files Game (HyperBole Studios
1998) the player could select the PCʼs mood for talking with NPCs. In The X-Files Game, these
were emotions like “humorous,” “indifference,” and “paranoia” that were presented graphically to
the player when talking to NPCs that the player clicked on to make the PC respond in that mood.
In Blade Runner the player must use the options menu to change the PCʼs mood which can alter
how the PC converses with NPCs. In both of these examples, the player has no control over
what the PC says, merely the mood in which the PC speaks in. This style of emotion control by
selecting the mood for a response often is just changing the “flavor” of the dialogue or it serves
as a means to an end for conversational puzzle solving. This system of emotion control has
several problems: lack of control over PC dialogue hurts role-playing, allowing the PC to go from
11
one mood to another can result in extreme mood swings that makes the PC appear
schizophrenic, and the effects of the mood choices to the game are unclear or negligible.
Affecting NPC Emotions
Most games with emotion input fall into the category of players affecting NPC emotions. In these
games, NPCs are given some semblance of having emotions to make them more dynamic,
interesting, and responsive. The playerʼs actions then affect or sometimes directly control the
NPCʼs emotional state, which can influence the game in a variety of ways.
The Sims (Maxis 2000) gives the player godlike control over the simulated people and their lives,
that brings them happiness or despair. While in the game this usually translates as a willingness
to obey the playerʼs commands -- and breaking into sobs if they are too unhappy -- the game is
very good at letting the player create their own story and thereby project emotions onto the
characters.
Ruben & Lullaby (Opertoon 2009) is an art game for the Apple iPhone that features a nonverbal
fight between a couple. The player can direct the argument by shaking the iPhone to make a
character mad or stroke the character on the touch-sensitive screen to calm them down. In this
manner, the player can get the characters to reconcile or break-up.
Façade (Procedural Arts 2005) is an interactive story, which also deals with a fighting couple.
The player, caught up in a feud between a married couple, can type in sentences to “converse”
with them. Then the natural-language processor deciphers it and it affects the couple and their
opinion of each other. In this manner the player can manipulate the couple, ultimately driving
them apart or having them reconcile. The only characterization the player is given is that they are
an old college friend of the coupleʼs and given no development. Besides being able to annoy the
12
couple enough to have them kick the PC out, the PCʼs relationship to the other characters is
never established or changed. This is supposed to allow the player to play themselves or
someone of their own imagination and develop their own story, so it is not necessarily a bad
thing. However, the stand-alone narrative does suffer because of this.
Wonder Project J (Almanic 1994) is an example of a “raising” type of game, where players “raise”
Pino, an android boy. Players use positive and negative reinforcement to train Pino in a variety of
skills and behavior, from petting (or kicking) a dog to throwing a ball correctly. Too much
punishment causes Pino to become angry with the player; praising him makes him happier with
the player. While the player can make Pino kind or cruel, a fighter or an intellectual, happy or
angry, etc., ultimately the player must force Pino to gain certain skills and acquire a certain
mind-set to pass game obstacles. Unfortunately, this means that players cannot raise Pino in any
way they would like; the training of Pino must conform to meet certain requirements for
progression through levels. For example, Pino must win a fighting challenge to proceed and
therefore must be trained to be a great warrior even if the player would prefer to make Pino
pacifist. There are many times when the player must unlearn skills or learn specific skills to pass
obstacles. In the end, this means that the emotional involvement between the player and Pino is
lessened as the player must view their management of Pino as a means to an end and not
become involved.
Visual novels, popular in Japan, are text-based games generally with still-image graphics and
whose player interaction is to select from multiple-choice decisions at specific points. Most of
these games, such as Clannad (Key 2004), focus on relationships between the PC and the other
characters. Player decisions can lead to characters hating or even loving the PC. Narratives in
visual novels general fall into learning about the NPCs, with the goal to get a player-chosen
character to love the PC.
13
Galatea (Short 2000) is a conversation-based, text-only interactive fiction (IF) game. Its premise
is to manipulate the NPC Galatea by asking her questions or telling her things. From this
interaction, her mood and opinion of the PC changes, which affects what she will tell the player
and how she does it, as well as resulting in several different endings.
By focusing on affecting NPC emotions, the narrative inherently becomes about the NPCs. The
characters having the emotions are driving the story, and without having a PC with emotions, the
PC becomes trivial if they are present at all -- merely a vessel for the player to act through. In
these types of games, the player orchestrates and guides the story, often not involved in the story
as a result.
Games with Emotion Output
Games with emotion output attempt to get the player themselves to experience emotions as a
result of playing the game. This must be directly related to something in the game and not just
being happy at finishing a level or angry at dying for the hundredth time. Rather, it is emotion as
a result of the narrative, as a result of gameplay, or as a result of choices.
Emotion as a Result of Narrative
The largest category of games with emotion output is those that evoke emotions as a result of
their narrative. This is often accomplished with a touching cutscene that borrows cinematic
techniques to manipulate the playerʼs emotions. Emotion can also be from what the gameʼs story
is about, especially if it is a “serious game” about emotionally volatile subjects.
Many story-based interactive fiction (IF) games fall into this category. For example, the game
Worlds Apart (Britton 1999) has a complex and involving story that builds an attachment to the
14
PCʼs mother as the player searches for her only to discover that she is dead. Photopia (Cadre
1998) tells a non-linear story about a car running over a girl, paying particular attention to people
in the girlʼs life so the player can feel their loss when she dies. Hunter, in Darkness (Plotkin 1999)
has the player hunting a terrible monster and at one point stuck in total darkness. A player is
quoted as saying about Hunter, in Darkness, “I was trapped in a cold, pitch-black, water-filled
crawl hole, and I was [expletive]-freaking scared” (Parker 2000).
There are also many video games that are narrative-driven and have emotional impact. In Final
Fantasy VII (Square 1997), a main party member, Aeris, dies and there is a touching funeral
cinematic. In Final Fantasy X (Square 2001), the PC Tidus sacrifices his existence to save the
world and the woman he loves. In the ending movie, Tidus is shown trying to hug Yuna although
he has become transparent and can no longer touch her. Tearfully, Yuna runs to him before he
completely disappears only to run through him. A fight between the PC and a young boy in
Xenogears (Square 1998) is also tangibly sad, as the PC had previously accidentally killed the
boyʼs sister, who was also the PCʼs best friend, on her wedding day. If the player does not attack
him, the boy will throw his sisterʼs wedding dress at the PC and run off in tears.
There is also emotion based on content, specifically in games that explore emotionally rich
subject matter. These are often identified as “serious games” because the message or topic of
the game carries emotional weight. Darfur is Dying (Ruiz 2006) is based on the real life genocide
in Darfur. A player plays as a refugee trying to survive. In the board game Train (Brathwaite
2009) players race to get their train of people to their destination -- which happens to be
Auschwitz, the largest Jewish concentration and death camp during the Holocaust.
Emotion as a result of narrative is extremely rich and varied, but many games utilize cinematic
techniques to evoke emotion. The ending of Final Fantasy X, for example, is extremely sad and
15
touching, but it happens in a cutscene, independent of player actions in the game. A better
technique would be to take advantage of the affordances of games, allowing for the player to
actively participate rather than passively observe. The game should involve the player directly
with the emotional aspect and tie it to player choice so that the player will feel responsibility for
the results. Player participation means they are more actively engaged and giving them
meaningful choices can result in the player investing their selves in the story. Game cutscenes
are very good at evoking emotion in players, but games should embrace the fact that they are
games and use player interactions to establish an emotional connection with the player.
Emotion as a Result of Gameplay
For emotion as a result of gameplay, the mechanics of the game itself evokes emotion in the
player. The prime example of this approach is found in the game Ico (Team Ico 2001), where the
player leads an NPC, Yorda, by the hand and must keep her safe. “Shadows” are after Yorda and
they will pull her down a shadow portal if they get the chance, resulting in a “game over.” When
Yorda is out of sight, an often occurrence, the player becomes anxious. When Yorda is being
pulled down a shadow portal, the player becomes afraid. The player is protective of Yorda,
although they know next to nothing about her since she speaks a different language.
Shadow of the Colossus (Team Ico 2005) requires the player to destroy rather than protect. The
goal of the game is to destroy the 16 colossi, who are beautiful, huge, and peaceful until the
player attacks them. The player feels sad killing the colossi because it is akin to committing
genocide. The game also fosters an attachment to the PCʼs horse, who follows the PC around
and who the player rides for transportation. Near the end, the horse is presumed dead and as a
result of the connection the player developed with the horse through interacting with the horse
creates a sense of palpable loss within the player.
16
In Cannon Fodder (Sensible Software 1993), the player commands soldiers to complete
objectives, such as kill all enemy soldiers or rescue hostages. Soldiers have unique names, and
when they die they are given gravestones on a hill that is displayed to the player before every
mission. At the end of each phase of a mission, soldiers that survive are awarded combat
bonuses and those that perished are replaced with new recruits. The player can have an
emotional attachment to the soldiers because they have names, they become more useful to the
player the longer they survive, and because their deaths have some consequence.
Using the actual gameplay mechanics to evoke emotions in players is highly successful but
extremely difficult to accomplish. Most of these games include a light narrative in order to focus
the player on the gameplay rather than the story. There is also no player choice involved, as the
game must force the player into a certain behavior to evoke emotions. In Ico, the player is forced
to protect Yorda because to do otherwise results in a “game-over.” The anxiousness that players
feel about Yorda stem from their desire to not get a “game-over” rather than from any perceived
care for Yordaʼs wellbeing. That does not make the emotion any less real, but it only exists
because the game forces the player into a specific behavior -- protecting Yorda. In Shadow of the
Colossus, the player is forced to kill the colossi to progress in the game. There is no choice in the
matter, and it is only because the player is forced to kill these creatures against the playerʼs will
that makes it sad. Players in Cannon Fodder must send soldiers to their deaths in order to win
the game no matter the playerʼs feelings in the matter or their attachment to the soldiers. To
generate these emotions in players, player choice is sacrificed. That is not to say this type of
emotion generation is worse than others, merely that to achieve it means the absence of player
choice. It also means that the emotions evoked are limited in their scope as this methodology
cannot support multiple emotions easily.
17
Emotion as a Result of Choice
Player choice facilitates player involvement by allowing the player to invest emotionally in the
characters and the world they inhabit. When given choices, a player is enabled to feel
responsibility for their actions. There are games that have lots of choices for the player to make,
games that have just a few pivotal choices to make, and games with no implicit choices offered
but the player has freedom of choice with their actions.
Games that offer many choices generally present these as moral choices, forcing the player to
choose to be “good” or “evil,” or to a lesser degree choosing to be nice or a jerk. Some examples
of these types of games are Star Wars: Knights of the Old Republic (BioWare 2003), Fable I & II
(Lionhead 2004, 2008), Mass Effect I & II (BioWare 2007, 2010), Dragon Age: Origins (BioWare
2009), Jade Empire (BioWare 2005), and Fallout 3 (Bethesda 2008). These games use choices
to present dilemmas to the player, and forcing responsibility for the result on the player as they
chose to do it. Their main failing is that by having the dichotomy between being “good” and “evil,”
the player needs only pick the choice that reflects the morality path they are following, which
reduces the responsibility aspect of player choice. Another common failing is the lack of
repercussions for their actions beyond the PC getting morality points. In Fallout 3 (Bethesda
2008), the player is asked to choose whether to kill a being at his own request or to keep him
alive for the benefit of humankind. A very emotional choice, but ultimately does not deliver, as
regardless of the choice made the other side will be accepting of the decision and no further
game repercussions are seen.
The Mass Effect series (BioWare 2007, 2010) is making strides towards choice repercussions by
having player choices affect the entire series. Choices made in the first game have some effect
on the game characters and world in the second. The choices in the first and second game will
also have repercussions in the upcoming third game, making the series the games with the
18
longest reaching repercussions. This adds incredible weight to player decisions, especially if the
long-term repercussions are not clear, and makes for a very responsive world and story that the
player will feel responsible for. There are also several examples of gray morality, where there is
no clear “good” or “evil” choice, such as giving the enemyʼs powerful technology to the less than
pure organization the player has been forced to work with. Giving them the technology will give
human forces an advantage in the coming war, but it has been shown that the organization has a
history of abusing technology and power, because they follow the credo of “the ends justify the
means.”
Most of these games are marginally successful at evoking emotion because most of these games
force the choices to be between doing the “good” option or the “evil” option. This makes the
choice almost pointless, and therefore engages the player less, which results in less emotional
impact. However, this is a relatively new genre and these games are getting more sophisticated
and subtle as they move away from the pure black and white world of “good” and “evil” and have
more substantial repercussions for player choices.
Games that have just a few decision points capitalize on there being serious repercussions for
their choice. The choice is also generally story related and not a “good” or “evil” binary choice. In
Suikoden II (Konami 1999), the player is forced to choose to slay the PCʼs best friend who
betrayed him or let him live. It is only the playerʼs feelings on the matter that determine what the
player is going to do, the repercussions are only story based. However, players who do not care
for the NPC will not find this an emotional choice. In Shin Megami Tensei: Nocturne (Atlus 2004),
the player gets the shape the new world by deciding which Reason they will follow. A Reason is a
philosophy of living that the new world will use as its foundation. The choices are between living
as “one” with everything and therefore destroying the concept of individuality, or of isolationism
where everyone lives in their own fabricated world, or survival of the fittest, where only the strong
19
will prosper. The player can also chose to have no Reason, which can result in the world
returning to the way it was or it can lead to the demons taking over. Because the player is
choosing what the new world will be, it is a weighty choice with serious repercussions. The
choice is also not a specific decision point where a list of options shows up. The player makes
this choice as they play, with factions belonging to each Reason trying to persuade them to side
with them and they all have good pros and cons to their Reason. In Nocturneʼs case, a lack of
direct correspondence between player action and the PCʼs Reason can make this choice feel
arbitrary.
Because of the scarcity and consequences of these choices, they become important and
emotionally laden. However, their focus on these choices can cause other player actions to be
seen as inconsequential. If these choices are not tied deeply with the story or if the story itself is
not emotional, they can also fail to be compelling.
Then there are games with so much freedom of choice that the player must take responsibility for
their actions as the game is not forcing them into a particular action. These games include
sand-box games like the Grand Theft Auto series outside of the main missions. In Grand Theft
Auto: Vice City (Rockstar North 2002), the player has immense freedom in what to do outside of
the missions. For example, they can use ambulances and save people, go on a killing spree, hire
hookers, raid the police station, or just play tourist and explore the city. However, the emotional
weight of being forced to take responsibility for their actions is countermanded by there being no
lasting consequence to these actions. The value of NPC lives is negligible as they are
represented as mere fodder, any police “heat” is removed with time or a new paint job on the
playerʼs car, and nobodyʼs opinion of the PC changes. A lack of consequence means there can
be no lasting emotional investment in the game.
20
Princess Maker 2 (Gainax 1993) is a similar, although more limited, game with freedom of choice.
In it, the player is gifted a child by the gods and has eight years to raise her as the player sees fit.
The player chooses jobs and school lessons for the girl to take and as such shapes her life.
When she becomes 18, the player is given one of over 60 endings about the future of their
“daughter” based on what stats the girl has, which are based on what the player has made her
do; have her work at the “Sleazy Bar” and she can become a mistress of a noble, have her work
at the church and she can become a nun, or spend some time having her chat with the prince
and she might become the next queen. However, the lack of any personality of the girl and the
sometimes seemingly arbitrary endings makes it difficult to care for the girl and therefore there is
little emotional involvement in the game. Cute Knight (Hanako Games 2006), an indie game, is
very similar but the player plays as the girl and they choose for themselves what they should do,
rather than playing as the “father” issuing orders.
Without clear and serious repercussions, a game with a high degree of freedom of choice cannot
capitalize on the emotional impact afforded by such freedom. Grand Theft Auto: Vice City lacks
repercussions and Princess Maker 2 lacks clarity of their repercussions.
21
Prior Art - Papers
There are two fields of study necessary for creating an emotionally responsive game. The game
must respond to emotion input and contain NPCs with well developed and comprehensive
emotion models that are able to respond to complex player actions. One interesting project for
emotion input used the playerʼs voice. The project used real-time tone analysis to interpret a
playerʼs voice for emotional content and used this to guide NPC responses (Cavazza et al. 2009).
A project about creating more realistic NPCs was concerned with creating an emotional
connection with NPCs by having them express their emotions visually. Mainly a tool for game
programmers, it allowed the user to change the behavior, body language, and mood of a
character (Nayak 2005). A body of work has been undertaken to create more believable virtual
human NPCs at USCʼs Institute of Creative Technology. They have explored how to make NPCs
become more realistic, changing their emotional state based on player action and to relay their
emotional state nonverbally (Marsella et al. 2001). To make believable virtual human NPCs, a lot
of work has been done to integrate NPCsʼ emotion and dialogue models (Traum et al. 2004),
including allowing an NPC to talk about their emotions (Muller et al. 2004). They are also working
on simulating “social emotions”, emotions that have a social component (like guilt), enabling
NPCs to assign blame or credit for better social interactions with the PC (Gratch et al. 2006).
To have an emotionally involving game, it is not only necessary to have an emotionally
responsive game but also to evoke emotions in the player. Janet Murray asks if games can
“make us cry,” meaning not to bring tears to the player, but to arouse emotions by having the
player identify with the characters and feel that their actions and events in the game have weight
and significance. Murray postulates that players do not particularly care about the people in
games save as targets or opponents, and events in the game as wins and losses, even if there is
22
a narrative present in the game (Murray 2005, 85). By not taking advantage of their inherent
interactive qualities, games do not have their storytelling integrated with the game format.
In order to maximize story power we have to integrate the interactorʼs actions
with the story content. When we are successful we create the experience of
dramatic agency, the cueing of the interactorʻs intentions, expectations, and
actions so that they mesh with the story events generated by the system. ...
User action is elicited and is always rewarded with an immediate effect that is
consistent with and progresses the story.
-- (Murray 2005, 85)
While games have been found to elicit emotions, happiness was difficult and sadness impossible
to be evoked outside of non-interactive cut-scenes using cinematic methods (Zagalo 2005, 112).
This is a failure of game design considering that by involving the player directly, they should be
more immersed in the story and therefore more affected by the events that occur. Non-
interactivity takes them out of the game, reminding them that it is just a scripted story. This is why
it is important to find interactive methods to emotionally involve the player.
To have the player be really emotionally involved, the player has to be given the ability to make
meaningful choices. To this end, the distinction between choices and problems are of particular
interest. In a recent article in Gamasutra, James Portnow describes the difference between
choices and problems:
Any decision that has a definitive best answer is a problem and not a choice.
Anything with a ʻsolutionʼ is a problem and not a choice. Anything that is a
calculation is a problem and not a choice. ... There is very little “free will”
involved in these types of problems, for the most part you know what you are
trying to execute ... and the challenge is in the execution. Anytime where you
present the player with a clearly defined goal and the “fun” or challenge is in the
execution of that goal you have presented them with a problem.
Choice appears when you are asked to decide between two things of equivalent
or incomparable value. ... Choices tend to be much harder to design than
problems because they donʼt have a clear right answer. They also tend to be
much harder to fit into games because their divergent nature tends to broaden
the scope of games. ...
Problems direct the player towards a goal; choices let the player choose their
goal.
-- (Portnow 2009)
23
To engross the player and engage them in a struggle with their choices, they need to be
presented with actual choices and not problems. That means that a reward system should not be
used unless it is strictly balanced so it does not become the overriding factor in the playerʼs
ultimate decision. Instead, the game must have a deep and interesting narrative that is integrated
with the playerʼs actions. The player must make all of the major decisions that the playerʼs
character requires. These decisions must be choices that are of equivalent value and have
meaningful consequences and not be mere problems.
24
User Testing and Evaluation
Throughout the development of Crimson Night users have been brought in to play the game to
provide feedback. Most play-testing was informal, with the testers giving feedback (“stream of
consciousness” methodology) as they played while notes were taken of their thoughts and
opinions. Later play-testing made use of screen-capture and video-recording of the player as well
as storing the playerʼs moodpath. User feedback was instrumental in tweaking the user interface
and game mechanics to make the game clearer and to refine how it uses emotion input.
“Moodpath,” in this paper, refers to the playerʼs set of emotion choices made in the game. It
represents the playerʼs journey or “path” through the game, which is based on the emotion or
“mood” choices made -- the path of the PCʼs mood through the game, in other words. Different
moodpaths can result in different endings, depending on how divergent the moodpaths are.
Moodpaths in this manner are used for play-testing, as they are recorded during play and then
used as a tool for analyzing player choices. However, moodpaths are also used to describe the
play experience itself. If a player is mostly choosing the “anger” options, then they would be
following the “anger moodpath.” Likewise, if they choose mostly sad options, they would be
following the “sad moodpath.” Players who choose generally one option in this manner follow a
“simple moodpath” while a player who uses many different emotions would be following a
“complex moodpath.”
Mechanics
Three main mechanics areas were revised based on user feedback: the conversation system, the
action system, and the mood change system. This was an iterative process where play-testersʼ
feedback was evaluated and changes were implemented. Each system was evaluated for
25
playability and emotional impact with successful elements kept and detrimental ones modified or
removed.
Mechanics - The Conversation System
In the first prototype, the conversation system had the player select the mood of the PCʼs
responses. While one player enjoyed the mystery of what the PC would say, most disliked the
lack of control over the dialogue. Another problem was the ability to jump from one mood to
another, leading players to view the PC as schizophrenic. To address the second problem of it
being too easy to have the PC suffer from rapid mood swings, each mood sans neutral were
divided into three levels of mild, moderate, and extreme for a total of 13 distinct moods. The first
revision of the system had the dialogue choices given in whatever mood the PC was in when the
conversation was initiated with no way to change her mood inside of the conversation. Players
were not interested in the conversation and it was detrimental to the goal of having emotion play
a key role in game interactions. Therefore, a new conversation system was developed that
combined mood choice and dialogue choice to satisfy playersʼ desire to choose what the PC
would say while still incorporating the goal of using emotion.
The first version of this new system had a set of responses and allowed the player to dynamically
change the mood that the PC would say these responses in. The moods available for the player
to choose from were the moods surrounding the current mood. So when at neutral, the player
could choose from neutral or any of the 4 mild emotions. At moderate anger, they could choose
from mild, moderate, or extreme anger. If the PCʼs mood was located in-between two moods,
then a combination of those two moodsʼ surrounding moods was used. This resulted in players
getting frustrated with going through all the possible moods to select the one they wanted as well
as making them feel that the mood did not matter as the mood was relegated to “flavoring” the
text. It also made it difficult to compare responses in different moods, as players had to switch
26
back and forth to view them. The mood choice system restricted the player too much as well,
making it so that once the player made a few choices they almost always got restricted to just
moving up and down in the level of a specific mood. For example, if the PC was at mild joy the
player could only choose from neutral, mild joy, or moderate joy, keeping them from deciding to
answer the NPC with any of the other emotions.
The next iteration on the conversation system featured a set of dialogue choices that reflected the
available mood choices of the player (Figure 4). These were not just mood “flavored” lines, but
were rather different responses based on the different moods. The moods available to the player
were based on the same system as before, but it kept the mild versions of the other moods (save
the one opposite of the PCʼs current mood) available. Now when the PC was at mild joy, the
player had neutral, mild joy, moderate joy, mild fear, and mild anger to choose from. Mild
sadness was not an option as on the mood spectrum, it is opposite of joy. This met with much
Figure 3: Screenshot of the first prototype, which was purely
conversational with only mood control for the mechanics
27
more success with the players, who enjoyed picking different responses. However, the absence
of the opposite emotion and the fact that choosing neutral made the PCʼs mood go towards
neutral was confusing or not obvious to players. Therefore the opposite emotionʼs mild response
was added to the list of choices and choosing neutral would make no change on the PCʼs mood.
As a result of work done for the emotion control system, a new conversation system was added
allowing the player to have the PC talk to herself about topics as well as use those topics in
conversation with NPCs. There became two conversation types -- active and responsive. In an
active conversation, the player takes the initiative and asks the NPC about topics. In a
responsive conversation, the NPC takes the initiative to ask the PC questions with the player
responding.
Mechanics - The Action System
Initially, there was nothing for the player to do outside of the conversation because the first
prototype was purely conversation based. This was uninteresting and did not make full use of the
Figure 4: Screenshot of a conversation in Crimson Night
28
goal of using emotions to tell a story and explore a world. Therefore, objects and NPCs were
given descriptions that changed with the PCʼs mood. This was met with wide approval and
interest from play testers. However, the method of accessing these descriptions had to go
through several iterations until the method of looking at them was pleasing to the players.
Initially, the player had to click somewhere to bring up an “action-wheel” (Figure 5) and then click
on “walk-to” to walk to that position, “look-at” to look at that object, or “talk-to” to talk to an NPC.
“Talk-to” was only available when clicking on NPCs, and “look-at” for clicking on objects and
NPCs, while “walk-to” was available on any clicked location. This was annoying to players as
they had to do two clicks to do anything, while also frustrating as it took too long to do what they
wanted. When looking at half a dozen objects, the repetition and time delay in seeing the
description was painful.
Figure 5: Close-up of the “action-wheel” used in an early prototype
29
The next iteration removed the “action-wheel” and instead used mouse click to mean “walk-to” a
location or “talk-to” if clicked on an NPC and mouse-over to be “look-at.” This was also frustrating
to players, as it was not clear that they had to hover the mouse over an object for a moment to
get the description. Many players became excessively frustrated and starting clicking randomly
trying to get something to happen, resulting in just the PC moving about. Even once they were
told the “look-at” mechanic used mouse-over, it was still annoying to the player because of the
delay in rolling over the object and receiving the description. The delay was necessary to
differentiate between a mouse-over for a “look-at” and an incidental mouse-over while on the way
to something else.
A new “look-at” mechanic needed to be developed, as it was one of the key mechanics for the
game and should be responsive and easy to use. Unfortunately, Flash, the development
environment, does not allow right clicks to be used so a simple left-click/right-click control scheme
for walking, talking, and looking could not be used. Instead, the mouse-over mechanic was kept
but the delay removed. To keep the game from being confused between a mouse-over for
“look-at” and an incidental one, the “shift” key must be pressed to enter “look-at” mode. Holding
down the “shift” and moving the mouse over an object of interest then would give its description.
Players found this easy to use, although they had to be told what to do. Initially, it was desired to
have only the mouse be used to interact with the game to keep it simple for the player. However,
the responsiveness of the game added by using the keyboard to augment the mouse control was
worth the added interface device.
Mechanics - The Mood Change System
As referred to in “Mechanics - The Conversation System” the player initially changed the PCʼs
mood by selecting which mood to respond with in a conversation. Subsequently, the player
directly controlled the PCʼs mood. The next version used a “mood-wheel” (Figure 6) with a little
30
face icon (Figure 7) to denote where the PCʼs mood and its location in the mood-spectrum.
Players could move this icon to change the PCʼs mood.
This direct control was annoying to the players, who had to manually set it, but also destroyed
any hope of fostering a relationship between player action and PC mood in a story-based way. It
was planned to use this system only when responding to something, so that the player could not
alter the PCʼs mood whenever they felt like it. However, early testing showed that even then it
would not work. The “mood-wheel” was distracting on the screen, the face-icon unclear, and the
mechanic of changing the mood too repetitious and story-breaking. Players unanimously disliked
the system, even if they enjoyed the results of changing the PCʼs emotion.
Figure 6: A screenshot of an early prototype identifying the “mood-wheel”
31
The next iteration had the PCʼs mood changed by what they looked at. Looking at a family
picture would make the PC happy, while looking at the bed would make them sad as the PC
pondered her lost dreams. Players did not easily pick up on the fact that looking at different
objects had different results and the emotion control became too muddied and uninteresting to
players.
Emotion changes were then facilitated in conversations, with the playerʼs dialogue choices
affecting the PCʼs mood. This was successful because the player felt they affected the PCʼs
mood and the connection between player choice and the resulting mood change were clear.
However, this meant that outside of conversations there was no mood change and the PC
became static.
To fix the problem of a mood-static PC outside of conversations, several additions were added.
The player could have the PC pace around to let the PCʼs mood move towards neutral. Players
Figure 7: A table of the moods displayed by the face-icon to denote
the PCʼs current mood in the “mood-wheel” used in an early prototype
32
did not find this intuitive or clear and it could take a bit of time to have the PC reach a neutral
mood. This was then changed to a “meditate” command. This left how to change the PCʼs mood
to a non-neutral mood outside of conversations. Not wanting to return to the failed mechanic of
looking at objects to change the PCʼs mood, the same “meditate” action was used, but
augmented by meditating on a topic. Topics had stored emotion tied to them, so meditating on a
topic would increase the PCʼs mood in the mood of the topic. This was an interesting mechanic,
but the player had no control over what mood was stored in a topic. A “monologue” system was
added, having the PC talk to herself about the topic as she meditated on it and using the same
conversation system used with other NPCs the player could alter the mood stored in the topic
along with the PCʼs own mood. These topics could then be used in active conversations with
NPCs, the PC talking about these topics in a mood based on the combined PCʼs current mood
and the mood stored in the topic.
Art
The sprite of the PC, as well as the NPCs to a lesser degree, changes based on the PCʼs mood.
This started with just a change in facial expression, but players often failed to notice this. The
first change was to bring the sprites closer when in a conversation, so the faces would be large
enough to recognize. This was still too subtle, making mood changes -- especially between
levels of an emotion -- too hard for the players to pick up on. Emotional “poses” that took into
account facial expression as well as body posture were utilized instead. Players could then easily
discern that the PCʼs emotions had changed to reflect her mood as a result of the playerʼs
choices.
33
However, outside of conversations, the sprite is not so large as to make this very noticeable.
There was also a focus on the PC being the only thing that changed, while the game was trying to
show how the world changed based on the PCʼs perspective due to her mood. For example,
object descriptions changed based on how the PC viewed that object while in her current mood.
There needed to be a visual tie in to the PCʼs perspective changing, so screen-effects based on
the PCʼs mood were added. The screen would turn redder as she got angrier, brighter when she
became happier, grayer as she got sadder, and darker and harder to see as she got more fearful
(Figure 8). Players enjoyed the screen effects, although some effects, like those for joy and fear,
needed additional iterations to display and interpret correctly. By utilizing player feedback and
iteratively implementing changes to the user interface, it has made it easier for players to
perceive the PCʼs current mood, whether inside or outside of conversations.
Figure 8: Mood screen effects. Center: Neutral, Clockwise from top: Extreme Joy,
Extreme Anger, Extreme Sadness, Extreme Fear
34
The PCʼs mood is altered based on the conversation choices. The PCʼs current mood is a
summation of all the mood choices the player has made. The PC sprite reflected her mood to
help convey this information to the player. However, this made players confused when the PC
was sad and they chose to have the PC say something angrily, and the PCʼs sprite remained
sad-looking. Therefore, when the PC “says” the playerʼs dialogue choice, the PC sprite was
changed to reflect the mood the response was in, and then reverted back to displaying the PCʼs
overall mood when the dialogue was completed. Players then understood that the PC was
“saying” the line in the mood they selected. Returning to the PCʼs current mood helped inform
the player that the PC had a persistent mood and was not just in the latest conversation mood.
Story
The story went through several changes as the project progressed. Initially, it was the story of
Alin Britemore, who, having been told that she has suffered a severe injury and might have
memory damage, tries to figure out who she really is. This was deemed too complicated, and the
story was changed to be that of a woman getting ready for an arranged marriage. This read very
well with players, who had no trouble understanding what was going on, but it was also viewed as
being too melodramatic. Play testers who had played the previous versionʼs story voiced their
preference for that one. The story was then restored to the previous adventure involving Alin.
As the game was developed, players began to voice confusion over the story setup as well as
desiring to have an actual ending. The game had been developed as a test of mechanics and not
as a full game, but this was detrimental to the testing. Therefore, a text intro and ending were
created. Initially, there were 5 different endings based on what the PCʼs overall mood was when
the game ended. The intro and endings were well received and no one had any trouble
understanding the setup for the story. Players also enjoyed that the ending changed as a result
of their actions.
35
A few players, reading an early prototypeʼs intro, said that they thought “Alin” was pronounced
“Allen” and therefore would be male, leading to some confusion when game play started and the
PC was female. To help alleviate confusion to the pronunciation of the name and gender of the
PC, the name was changed to “Alyn” (“A-Lynn”).
One prototypeʼs game play revolved around a psychiatrist visiting the PC in her room to introduce
himself and have their first therapy session. Many players thought that a man coming into a
womanʼs room was weird, and were put off by this scenario. Therefore, this scenario was
removed from subsequent prototypes and instead the player converses with the PCʼs parents in
her room. The location remained the PCʼs room as it has the most relevant objects to the PC and
can make the best use of the emotion system.
36
Discussion
Because this is relatively new design territory, there was a lot of trial and error involved in figuring
out a story to support the design goals as well as the critical emotional control structure. User
testing helped to improve all these facets, although there is still much work to be done.
According to the results from player tests, Crimson Night has succeeded in creating interesting
gameplay that offers a new experience to players, even through the level of emotional
involvement through manipulation of the PCʼs emotions was not achieved to the desired extent.
However, even marginal involvement would support the hypothesis that emotion input is
important to the future of game design by creating emotionally responsive games. The fact that
people enjoyed the game and played it multiple times shows more than marginal player
involvement and thereby supports the premise of this thesis. Nevertheless, there is a lot of room
for improvement here. A more elegant and natural mechanic for manipulating the PCʼs emotions
needs to be developed and this mechanic and the story needs to be integrated into a coherent
whole. More player actions based on the PCʼs emotional state need to be added as well,
particularly actions like hugging, holding hands, hitting, cowering, crying, etc. The PCʼs current
mood and the changes in the PCʼs mood need to be made clearer. NPCs need to have a more
detailed and complex emotion simulation to make them respond in a more believable manner.
Additionally, it is necessary to have several dilemmas of a non-trivial nature that is guided by the
PCʼs emotional state. It may also be beneficial for the user interface to move away from the
mouse and keyboard and find something more intuitive for the player and perhaps tied in to the
gameplay.
37
The goal of emotion input is to have the game world, comprising of the environment, characters,
and events of the game, be emotionally responsive to the player. Crimson Night achieved this on
a small scale. It is anticipated that others building on its foundation will be able to make a game
that pays attention not only to player actions but also to the playerʼs emotions (or at least their
choice of emotions). It would be interesting to see a game where the PC can erupt into
understandable tears and have an NPC comfort them or turn a cold shoulder; or a game where
the PC can kill someone in psychopathic glee or in the grips of terror; or a game that is aware of
and reacts to the PCʼs personality and changing emotions. These games would, in essence,
mold themselves to emotional data in conjunction with player actions to make a unique, custom
tailored game world that would be most satisfying.
38
Conclusion
Games are becoming more sophisticated and are getting closer to one of the key goals of
creating play that truly engages the player emotionally by establishing an emotion dialogue with
the player. While corporate game companies focus on the emotion output side of this with
morality games and emotional narrative games, this thesis argues that the neglect of the emotion
input side is detrimental to obtaining this goal. The thesis project, Crimson Night, was an
interesting game and its modest success in involving the player suggests the importance and
usefulness of designing for emotion input.
This thesis project shows that emotion input is a viable source of emotional involvement for the
player. This method working in tandem with established emotion output methods could
eventually create a genuinely emotionally involving and emotionally responsive game. This
would be a game that would evoke emotions in the player and incorporate emotional feedback
from the player. It is believed that only through this feedback loop that real emotional immersion
lies. Continued work on developing and refining emotion input methodologies will achieve the
goal of creating a truly emotionally responsive game.
39
Sorted Bibliography
Papers
Cavazza, Marc, David Pizzi, Fred Charles, Thurid Vogt, and Elisabeth André. “Emotional Input
for Character-based Interactive Storytelling.” Proceedings of the 8th International
Conference on Autonomous Agents and Multiagent Systems (AAMAS 2009) (2009):
313-320. (http://www.aamas-conference.org/Proceedings/aamas09/pdf/01_Full
%20Papers/05_28_FP_0842.pdf)
Descartes, René. The Passions of the Soul. Translated into English by unknown author. Printed
for A.C. and to be sold by J. Martin and J. Ridley. London, 1650. (found online at http://
net.cgu.edu/philosophy/descartes/Passions_Letters.html)
Ekman, Paul. “Universals And Cultural Differences In Facial Expressions of Emotions.” In
Nebraska Symposium on Motivation, ed. J. Cole, pp. 207-283. Lincoln, Neb.: University
of Nebraska Press, 1972.
Gratch, Jonathan, Wenji Mao, and Stacy Marsella. “Modeling Social Emotions and Social
Attributions.” In Cognition and Multi-Agent Interactiong: Extending Cognitive Modeling to
Social Simulation, edited by R. Sun, 219-251. Cambridge: Cambridge University Press,
2006. (http://ict.usc.edu/files/publications/Chapter06_gratch.pdf)
Jensen, Anabel. In an interview with Joshua Freedman, “What are emotions?” Six Seconds, June
2000. http://www.6seconds.org/modules.php?name=News&file=article&sid=316
(accessed March 28th, 2010)
Johnson, Gregory. “Theories of Emotion.” The Internet Encyclopedia of Philosophy. June 2009.
http://www.iep.utm.edu/emotion/ (accessed March 27th, 2010)
Marsella, Stacy, Jonathan Gratch, and Jeff Rickel. "The Effect of Affect: Modeling the Impact of
Emotional State on the Behavior of Interactive Virtual Humans." In Agents 2001
Workshop on Representing, Annotating, and Evaluating Non-Verbal and Verbal
Communicative Acts to Achieve Contextual Embodied Agents, Montreal, Canada, June
2001 (http://ict.usc.edu/files/publications/Agents-01-workshop-Marsella.pdf)
Muller, T.J., A. Hartholt, S. Marsella, J. Gratch, and D. Traum. “Do You Want To Talk About It?”,
Lecture Notes in Computer Science, Affective Dialogue Systems, Tutorial and Research
Workshop, ADS 2004, June 2004. pp. 188-192. (http://ict.usc.edu/files/publications/
dywttai5.pdf)
Murray, Janet, “Did It Make You Cry? Creating Dramatic Agency in Immersive Environments”,
Lecture Notes in Computer Science, Virtual Storytelling, November 2005. pp. 83-94.
40
Nayak, Vishal. “Emotional Expressiveness through the Body Language of Characters in
Interactive Game Environments.” MS thesis, University of California, Santa Barbara,
2005. (http://ilab.cs.ucsb.edu/publications/NayakMS.pdf)
Parker, Marnie. “The Meta-Puzzle of Interactive Fiction: Why We Like What We Like: Interactivity
Is The Message.” http://www.iffydoemain.org/theory/introduction.html (accessed March 1,
2010)
Plutchik, Robert. “The Nature of Emotions.” American Scientist 89 (2001): pp. 344 - 350.
Portnow, James. “Opinion: The Problem of Choice.” Gamasutra, http://gamasutra.com/php-bin/
news_index.php?story=22456, April 2009. (accessed February 1, 2010)
Spinoza, Baruch. Ethics. Translated from the 1677 Latin version into English by R. H. M. Elwes in
1883. Put online by Ron Bombardi. Department of Philosophy, Middle Tennessee State
University, 1997. http://frank.mtsu.edu/~rbombard/RB/Spinoza/ethica-front.html
(accessed March 27th, 2010)
Traum, David, Stacy Marsella, and Jonathan Gratch. “Emotion and Dialogue in the MRE Virtual
Humans.” Lecture Notes in Computer Science, Affective Dialogue Systems, Tutorial and
Research Workshop, ADS 2004, June 2004 pp. 117- 127 (http://ict.usc.edu/files/
publications/ads3.pdf)
Zagalo, N., A. Torres, and V. Branco. “Emotional Spectrum developed by Virtual Storytelling.”
Lecture Notes in Computer Science, Virtual Storytelling, November 2005. pp. 105-114.
Games
Almanic. Wonder Project J, Enix, 1994. (SNES)
Atlus. Shin Megami Tensei: Nocturne, Atlus, 2004. (PS2)
Bethesda Game Studios. Fallout 3, Bethesda Softworks, 2008. (PS3, Xbox 360)
Brathwaite, Brenda. Train, 2009. Information taken from “TGC 2009: How a Board Game Can
Make You Cry” written by Jordan Deam, The Escapist. http://
www.escapistmagazine.com/articles/view/conferences/tgc_2009/6021-TGC-2009-How-a-
Board-Game-Can-Make-You-Cry (accessed February 10, 2010) (Board game)
Britton, Suzanne. Worlds Apart, Bafʼs Guide to the Interactive Fiction Archive Archive, 1999.
http://wurb.com/if/game/403 (accessed March 1, 2010) (IF)
BioWare. Dragon Age: Origins, Electronic Arts, 2009. (Xbox 360, PS3, PC, MAC)
BioWare. Jade Empire, Microsoft Game Studios, 2005. (Xbox, PC)
41
BioWare. Mass Effect, Microsoft Game Studios, 2007. (Xbox 360)
BioWare. Mass Effect 2, Electronic Arts, 2010. (Xbox 360)
BioWare. Star Wars: Knights of the Old Republic, LucasArts, 2003. (Xbox, PC)
Cadre, Adam. Photopia, Bafʼs Guide to the Interactive Fiction Archive, 1998. http://wurb.com/if/
game/255 (accessed March 1, 2010) (IF)
Capcom. Breath of Fire III, Capcom, 1998. (PS1)
Capcom. Breath of Fire IV, Capcom, 2000. (PS1)
Cecropia. The Act, Cecropia, 2007. (Arcade, http://theactgame.com/)
Gainax. Princess Maker 2, Gainax, 1993. (PC)
Hanako Games. Cute Knight, iWin, 2006. (PC)
HyperBole Studios. The X-Files Game, Fox Interactive, 1998. (MAC, PC)
Key. Clannad, Visual Artʼs, 2004. (PC)
Konami. Suikoden II, Konami, 1999. (PS1)
Lionhead Studios. Fable, Microsoft Game Studios, 2004. (Xbox, PC)
Lionhead Studios. Fable II, Microsoft Game Studios, 2008. (Xbox 360)
Maxis. The Sims, Electronic Arts, 2000. (PC)
Nintendo. Super Princess Peach, Nintendo, 2006. (Nintendo DS)
Opertoon. Ruben & Lullaby, Apple App Store, 2009. (Apple iPhone)
Plotkin, Andrew. Hunter, in Darkness, Bafʼs Guide to the Interactive Fiction Archive, 1999. http://
wurb.com/if/game/483 (accessed March 1, 2010) (IF)
Procedural Arts. Façade, Procedural Arts, 2005. http://www.interactivestory.net/ (PC)
Rockstar North. Grand Theft Auto: Vice City, Rockstar Games, 2002. (PS2, PC)
Ruiz, Susana. Darfur is Dying, mtvU, 2006. http://www.darfurisdying.com/ (Web)
Sensible Software. Cannon Fodder, Virgin Interactive, 1993. (Amiga)
Short, Emily. Galatea, Bafʼs Guide to the Interactive Fiction Archive, 2000. http://wurb.com/if/
game/1326 (accessed March 1, 2010) (IF)
Square. Final Fantasy VII, Sony Computer Entertainment, 1997. (PS1)
42
Square. Final Fantasy X, Square EA, 2001. (PS2)
Square. Xenogears, Square EA, 1998. (PS1)
Team Ico. Ico, Sony Computer Entertainment, 2001. (PS2)
Team Ico. Shadow of the Colossus, Sony Computer Entertainment, 2005. (PS2)
Westwood Studios. Blade Runner, Virgin Interactive Entertainment, 1997. (PC)
43
Comprehensive Bibliography
Almanic. Wonder Project J, Enix, 1994. (SNES)
Atlus. Shin Megami Tensei: Nocturne, Atlus, 2004. (PS2)
Bethesda Game Studios. Fallout 3, Bethesda Softworks, 2008. (PS3, Xbox 360)
Brathwaite, Brenda. Train, 2009. Information taken from “TGC 2009: How a Board Game Can
Make You Cry” written by Jordan Deam, The Escapist. http://
www.escapistmagazine.com/articles/view/conferences/tgc_2009/6021-TGC-2009-How-a-
Board-Game-Can-Make-You-Cry (accessed February 10, 2010) (Board game)
Britton, Suzanne. Worlds Apart, Bafʼs Guide to the Interactive Fiction Archive Archive, 1999.
http://wurb.com/if/game/403 (accessed March 1, 2010) (IF)
BioWare. Dragon Age: Origins, Electronic Arts, 2009. (Xbox 360, PS3, PC, MAC)
BioWare. Jade Empire, Microsoft Game Studios, 2005. (Xbox, PC)
BioWare. Mass Effect, Microsoft Game Studios, 2007. (Xbox 360)
BioWare. Mass Effect 2, Electronic Arts, 2010. (Xbox 360)
BioWare. Star Wars: Knights of the Old Republic, LucasArts, 2003. (Xbox, PC)
Cadre, Adam. Photopia, Bafʼs Guide to the Interactive Fiction Archive, 1998. http://wurb.com/if/
game/255 (accessed March 1, 2010) (IF)
Capcom. Breath of Fire III, Capcom, 1998. (PS1)
Capcom. Breath of Fire IV, Capcom, 2000. (PS1)
Cavazza, Marc, David Pizzi, Fred Charles, Thurid Vogt, and Elisabeth André. “Emotional Input
for Character-based Interactive Storytelling.” Proceedings of the 8th International
Conference on Autonomous Agents and Multiagent Systems (AAMAS 2009) (2009):
313-320. (http://www.aamas-conference.org/Proceedings/aamas09/pdf/01_Full
%20Papers/05_28_FP_0842.pdf)
Cecropia. The Act, Cecropia, 2007. (Arcade, http://theactgame.com/)
Descartes, René. The Passions of the Soul. Translated into English by unknown author. Printed
for A.C. and to be sold by J. Martin and J. Ridley. London, 1650. (found online at http://
net.cgu.edu/philosophy/descartes/Passions_Letters.html)
44
Ekman, Paul. “Universals And Cultural Differences In Facial Expressions of Emotions.” In
Nebraska Symposium on Motivation, ed. J. Cole, pp. 207-283. Lincoln, Neb.: University
of Nebraska Press, 1972.
Gainax. Princess Maker 2, Gainax, 1993. (PC)
Gratch, Jonathan, Wenji Mao, and Stacy Marsella. “Modeling Social Emotions and Social
Attributions.” In Cognition and Multi-Agent Interactiong: Extending Cognitive Modeling to
Social Simulation, edited by R. Sun, 219-251. Cambridge: Cambridge University Press,
2006. (http://ict.usc.edu/files/publications/Chapter06_gratch.pdf)
Hanako Games. Cute Knight, iWin, 2006. (PC)
HyperBole Studios. The X-Files Game, Fox Interactive, 1998. (MAC, PC)
Jensen, Anabel. In an interview with Joshua Freedman, “What are emotions?” Six Seconds, June
2000. http://www.6seconds.org/modules.php?name=News&file=article&sid=316
(accessed March 28th, 2010)
Johnson, Gregory. “Theories of Emotion.” The Internet Encyclopedia of Philosophy. June 2009.
http://www.iep.utm.edu/emotion/ (accessed March 27th, 2010)
Key. Clannad, Visual Artʼs, 2004. (PC)
Konami. Suikoden II, Konami, 1999. (PS1)
Lionhead Studios. Fable, Microsoft Game Studios, 2004. (Xbox, PC)
Lionhead Studios. Fable II, Microsoft Game Studios, 2008. (Xbox 360)
Marsella, Stacy, Jonathan Gratch, and Jeff Rickel. "The Effect of Affect: Modeling the Impact of
Emotional State on the Behavior of Interactive Virtual Humans." In Agents 2001
Workshop on Representing, Annotating, and Evaluating Non-Verbal and Verbal
Communicative Acts to Achieve Contextual Embodied Agents, Montreal, Canada, June
2001 (http://ict.usc.edu/files/publications/Agents-01-workshop-Marsella.pdf)
Maxis. The Sims, Electronic Arts, 2000. (PC)
Muller, T.J., A. Hartholt, S. Marsella, J. Gratch, and D. Traum. “Do You Want To Talk About It?”,
Lecture Notes in Computer Science, Affective Dialogue Systems, Tutorial and Research
Workshop, ADS 2004, June 2004. pp. 188-192. (http://ict.usc.edu/files/publications/
dywttai5.pdf)
Murray, Janet, “Did It Make You Cry? Creating Dramatic Agency in Immersive Environments”,
Lecture Notes in Computer Science, Virtual Storytelling, November 2005. pp. 83-94.
Nayak, Vishal. “Emotional Expressiveness through the Body Language of Characters in
Interactive Game Environments.” MS thesis, University of California, Santa Barbara,
2005. (http://ilab.cs.ucsb.edu/publications/NayakMS.pdf)
Nintendo. Super Princess Peach, Nintendo, 2006. (Nintendo DS)
45
Opertoon. Ruben & Lullaby, Apple App Store, 2009. (Apple iPhone)
Parker, Marnie. “The Meta-Puzzle of Interactive Fiction: Why We Like What We Like: Interactivity
Is The Message.” http://www.iffydoemain.org/theory/introduction.html (accessed March 1,
2010)
Plotkin, Andrew. Hunter, in Darkness, Bafʼs Guide to the Interactive Fiction Archive, 1999. http://
wurb.com/if/game/483 (accessed March 1, 2010) (IF)
Plutchik, Robert. “The Nature of Emotions.” American Scientist 89 (2001): pp. 344 - 350.
Portnow, James. “Opinion: The Problem of Choice.” Gamasutra, http://gamasutra.com/php-bin/
news_index.php?story=22456, April 2009. (accessed February 1, 2010)
Procedural Arts. Façade, Procedural Arts, 2005. http://www.interactivestory.net/ (PC)
Rockstar North. Grand Theft Auto: Vice City, Rockstar Games, 2002. (PS2, PC)
Ruiz, Susana. Darfur is Dying, mtvU, 2006. http://www.darfurisdying.com/ (Web)
Sensible Software. Cannon Fodder, Virgin Interactive, 1993. (Amiga)
Short, Emily. Galatea, Bafʼs Guide to the Interactive Fiction Archive, 2000. http://wurb.com/if/
game/1326 (accessed March 1, 2010) (IF)
Spinoza, Baruch. Ethics. Translated from the 1677 Latin version into English by R. H. M. Elwes in
1883. Put online by Ron Bombardi. Department of Philosophy, Middle Tennessee State
University, 1997. http://frank.mtsu.edu/~rbombard/RB/Spinoza/ethica-front.html
(accessed March 27th, 2010)
Square. Final Fantasy VII, Sony Computer Entertainment, 1997. (PS1)
Square. Final Fantasy X, Square EA, 2001. (PS2)
Square. Xenogears, Square EA, 1998. (PS1)
Team Ico. Ico, Sony Computer Entertainment, 2001. (PS2)
Team Ico. Shadow of the Colossus, Sony Computer Entertainment, 2005. (PS2)
Traum, David, Stacy Marsella, and Jonathan Gratch. “Emotion and Dialogue in the MRE Virtual
Humans.” Lecture Notes in Computer Science, Affective Dialogue Systems, Tutorial and
Research Workshop, ADS 2004, June 2004 pp. 117- 127 (http://ict.usc.edu/files/
publications/ads3.pdf)
Zagalo, N., A. Torres, and V. Branco. “Emotional Spectrum developed by Virtual Storytelling.”
Lecture Notes in Computer Science, Virtual Storytelling, November 2005. pp. 105-114.
Westwood Studios. Blade Runner, Virgin Interactive Entertainment, 1997. (PC)
46
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Come with Me: a cooperative game focusing on player emotion
PDF
Pluff: creating intersections between tactile interface devices and fabric‐based electronics
PDF
Combiform: a console for the new communal casual game genre
PDF
Penrose Station: an exploration of presence, immersion, player identity, and the unsilent protagonist
PDF
There You Are: an exploration of storytelling methods using in video games
PDF
Players play: extending the lexicon of games and designing for player interaction
PDF
Courier: Dragons Within, a video game. An exploration into the magic circle as healing circle: restorative game design for a masculine framework free from the template of domination [document]
PDF
BeachcomberVR: inspiring exploration: concepts in the production of a virtual reality game
PDF
Ascension: an analysis of game design for speech recognition system usage and spatialized audio for virtual reality
PDF
Mac goes not back, but towards a more responsible representation of archaeology in video games
Asset Metadata
Creator
Wilcox, Brandi
(author)
Core Title
Emotion control of player characters: creating an emotionally responsive game
School
School of Cinematic Arts
Degree
Master of Fine Arts
Degree Program
Interactive Media
Publication Date
05/05/2010
Defense Date
03/25/2010
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
emotion control,emotion input,emotionally responsive game,emotions in games,interactive emotions,OAI-PMH Harvest
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Balsamo, Anne (
committee chair
), De Meo, Paul (
committee member
), Ellis, Mike (
committee member
), Goodsell, Joel (
committee member
), Hoberman, Perry (
committee member
)
Creator Email
asylia@hotmail.com,bwilcox@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m3010
Unique identifier
UC1471246
Identifier
etd-Wilcox-3514 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-330430 (legacy record id),usctheses-m3010 (legacy record id)
Legacy Identifier
etd-Wilcox-3514.pdf
Dmrecord
330430
Document Type
Thesis
Rights
Wilcox, Brandi
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
emotion control
emotion input
emotionally responsive game
emotions in games
interactive emotions