Ego-perspective Simulation Of Moral Dilemmata And Identification Of

Transcript

U NIVERSITY OF B AMBERG M ASTER ’ S T HESIS Ego-Perspective Simulation of Moral Dilemmata and Identification of Behavioural Predictors of Moral Decisions Author: Christian T EICHMANN (Matr. No.: 1760765) Supervisor: Prof. Dr. Ute S CHMID A thesis submitted in fulfilment of the requirements for the degree of Master of Science in Computing in the Humanities April 28, 2016 ii “You can discover more about a person in an hour of play than in a year of conversation. ” Plato (427 B.C. - 347 B.C.) “Beim Spielen kann man einen Menschen in einer Stunde besser kennenlernen, als im Gespräch in einem Jahr. ” Platon (427 v.Chr. - 347 v.Chr.) iii UNIVERSITY OF BAMBERG Abstract Faculty of Information Systems and Applied Computer Sciences Computing in the Humanities Master of Science Ego-Perspective Simulation of Moral Dilemmata and Identification of Behavioural Predictors of Moral Decisions by Christian T EICHMANN A situation without a “back door”, including simply negative decision options, followed by an unpredictable outcome with consequences we can not estimate - this could be a moral dilemma. Is there anything worse? Everyone might act differently facing moral dilemmata or do we have something in common? How do we ascertain our decision and is there an impact of the consequences on us? There are plenty of questions we could derive from dealing with moral dilemmata. Is it possible to get an answer for each of them? Researchers of different decades have dealt with this topic and have developed different methods to study their hypotheses. And with the technological development new tools can be used to carry out a study. Even new media like games can be helpful for research, but to which extend? Computer games are a very new medium if you consider all other media like films or books. They become very popular within a very short time. And not only they are popular among young adults but among all age groups. This aspect makes them attractive for the research in moral dilemmata. In this thesis I’d like to introduce the implementation of a simulation which will be part of a study from Ena Franz and Victoria Göetz, two psychology students in the master program. We have developed a conceptual basis for the program and cooperated during the implementation phase of the game. As a short introduction I’d like to present some previous work on the topic morality. It will be followed by the conceptual groundwork and the implementation of the simulation itself. At least I wanted to say a few words to the potential for the future research. v Acknowledgements I want to thank Professor Ute Schmid for the opportunity to write the masters thesis and also for the guidance and helpfulness. I also want to thank Ena Franz, Victoria Götz and Marius Raab from the department of general psychology and methodology for the great collaboration before and during the development of the game. vii Contents Abstract iii Acknowledgements v 1 Introduction 1 2 Previous Work on Morality - An Overview 2.1 Theoretical Traditions . . . . . . . . . . 2.1.1 Hume’s Emotionalist Tradition Moral Foundations Theory . . 2.1.2 Kantanian Rationalism . . . . . Stage Model . . . . . . . . . . . 2.1.3 Encounter . . . . . . . . . . . . Moral Dilemma . . . . . . . . . Moral Reaction . . . . . . . . . 2.1.4 Moral Emotions . . . . . . . . . 2.2 Morality in Games . . . . . . . . . . . 2.2.1 Mode of Interaction . . . . . . 2.2.2 Aspects of Player Behaviour . Players Behaviour by Age . . . 3 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 3 3 4 4 5 5 5 7 7 8 8 9 10 Moral Dilemmata and Conceptual Game Design 3.1 Theoretical Model . . . . . . . . . . . . . . . . 3.2 Building a Game . . . . . . . . . . . . . . . . 3.2.1 Environment and Interface . . . . . . 3.2.2 Character Features . . . . . . . . . . . Facial Expressions . . . . . . . . . . . Gesture & Full Body Movement . . . Methods . . . . . . . . . . . . . . . . . 3.2.3 Character Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 13 15 16 17 17 18 18 19 . . . . . . . . . . . . 21 21 22 22 22 24 26 27 29 29 29 30 30 Realisation of ‚Projekt: Rose‘ 4.1 Technical Details . . . . . . . . . 4.2 Character Creation . . . . . . . . 4.2.1 Human Body . . . . . . . 4.2.2 Clothes and Hair Creation 4.3 Character Animation . . . . . . . 4.4 Scenes . . . . . . . . . . . . . . . . 4.5 Interfaces . . . . . . . . . . . . . . 4.6 Functions . . . . . . . . . . . . . . 4.6.1 Interface Functions . . . . 4.6.2 Randomized Scenes . . . 4.6.3 Character Detection . . . 4.6.4 Dialogue System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii . . . . 31 32 33 33 Resume & Potential for Future Research 5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Proposal for Improvement . . . . . . . . . . . . . . . . . . . . 5.3 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . 35 35 35 36 4.7 5 4.6.5 Speech and Thought Bubbles 4.6.6 Data Collection . . . . . . . . 4.6.7 Scene Skipping . . . . . . . . Discussion on Predictability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bibliography 37 A Downloaded Assets 39 B Content of the DVD 41 C Scenes 43 ix List of Figures 3.1 3.2 3.3 3.4 3.5 Dialogue Dilemma 1 Dialogue Dilemma 2 Dialogue Dilemma 3 Dialogue Dilemma 4 Dialogue Dilemma 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 14 15 15 15 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 MakeHuman User Interface . . . . . . . . . . . Parameters for adjusting the hair. . . . . . . . . Keyframes within the animation screen layout. T-Pose . . . . . . . . . . . . . . . . . . . . . . . . Example for Animator Controller . . . . . . . . Start Interface . . . . . . . . . . . . . . . . . . . Ingame Screenshot . . . . . . . . . . . . . . . . Rose Life Cycle (from left to right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 24 25 26 27 28 28 32 C.1 C.2 C.3 C.4 C.5 Scene 1 Scene 2 Scene 3 Scene 4 Scene 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 44 44 45 45 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi List of Tables A.1 Downloaded Assets . . . . . . . . . . . . . . . . . . . . . . . . 39 xiii List of Abbreviations DDA DDE e.g. fMRI i.a. MFT MMO PvE PvP RPG SSGS Doctrine (of) Doing (and) Allowing Doctrine (of) Double Effect example given functional Magnetic Resonance Imaging inter alia (eng.: among other things) Moral Foundations Theory Massive Multiplayer Online Player versus Environment Player versus Player Role Play Game State Shame (and) Guilt Scale 1 Chapter 1 Introduction In our everyday life we are facing various situations we have to respond. They are substantially distinguished from each other. Some situations seem very simple and easy to manage (e.g. I’m thirsty, therefore I’ll drink something). Other ones might be more complex and difficult to handle. Thereby the complexity itself varies between humans, which is probably the result of our personality and the understanding of the situation we are confronted with. For one it is easy to make a decision in a given situation but another one can be difficult to make a decision. For example, there is an injured animal on the roadside of a busy street. How many people would recognize it or consider helping? I assume, that many people would pass it because they don’t notice it or feel it as a reason to stop for. But others may fall into an internal conflict and thought of different opportunities to help. Considerations like "Shall I hold on to help?" or "Should I call someone who is in charge for injured animals?" Such conflict which require a decision and independent of your choice the outcome will be negative at least for one party, can be described as a dilemma. Dilemmata are nothing new to us and everyone of us has at least met one, where we have to make a decision. And this decision might be not so easy to make. The difficulty of this case is not just the effect of our inner conflict we have to cope with. There is much more to it like we can’t imagine the dimension of the consequences it can cause. An interesting variant of such conflicts might be the moral dilemma. Because there is not just the inner conflict and the possible consequences we have to manage but the available decisions which exclude each other. A precise definition for moral dilemmata does not exist yet but this should not be discussed here. Until today there are only tendencies emerged from different approaches. Moral dilemmata are not exclusive to the real world. More and more video game producers tend to integrate moral dilemmata in games. How well this is implemented should not be the topic here. However, what does that mean and which impact does it have to all of the players? Is it possible to predict the behaviour of a player or which decision he or she will make? What advantage would we get, if we could predict it and which insights would we get? In the following chapter I’d like to deal with morality and moral dilemmata on a theoretical level. In addition, the medium game should be considered, too because it become more and more popular. Therefore it will be a good opportunity for research to attract players. In cooperation with to psychology students in the Master’s program we have developed a groundwork for a simulation where players have to face different moral dilemmata. In each dilemmata the player has to make a decision and irrespective 2 Chapter 1. Introduction of the decision the outcome will be negative. The implementation of the simulation was carried out by myself and through additional consultations with the psychology students we have adjust the simulation. Opportunities for the implementation and the final realization are explained within the main part. Afterwards I’d like to give an overview of the thesis as well as some proposals for the improvement of the simulation. At least a few words should be spend on the potential for further research. 3 Chapter 2 Previous Work on Morality An Overview 2.1 Theoretical Traditions In the past, two theoretical traditions based on the work of two philosopher’s studies evolved. The traditions of Hume and Kant are prominent until today because they have much influence on a wide range of researchers of different disciplines. Though I’d like to introduce both approaches focusing the moral judgement and including some modern extensions developed by researchers. 2.1.1 Hume’s Emotionalist Tradition Hume has hold his thoughts mainly in the two works “A Treatise of Human Nature” (a three-volume study, 1738) and afterwards “An Enquiry Concerning Human Understanding” (1748). These studies build the foundation of the so called “Emotionalism”. They lead to further studies containing his view of moral behaviour. The “Emotionalism” implies that reasoning is guided by emotions. And therefore emotions are the primary cause of moral judgement. Representatives (e.g. Jonathan Haidt) of this approach guess, that moral judgement is defined by strong emotional reactions. These emotions arrive in form of swift, affect-laden moral intuitions. Haidt assumes, that rational descriptions of individuals result from “post-hoc rationalizations”. More drastically, Haidt means that “respondents acknowledge, having no rational defence for their position” and called this “moral dumbfounding” (Monin, Pizarro, and Beer, 2007). In addition, “Emotionalism” state the hypothesis that the diversity of “cognitive shortcuts of individuals rely upon [their] cognitive capacities [which] are insufficient for the task” (Monin, Pizarro, and Beer, 2007). This indicates that “rationalizations” can’t be a main feature within moral reactions. Instead a “heuristic” approach of processing “enables people to function effectively where information is incomplete and time pressures or demands on their attention prevent [them] from applying full reasoning capacity” (Monin, Pizarro, and Beer, 2007). Therefore it is believed that several judgements and decisions apparently avoid deliberate considerations. And also the articulation of the reason of behaviour, “reveal people’s ability to generate narratives for their own post-hoc fashion, rather than demonstrating any direct access to the real causes of their choices” (Monin, Pizarro, and Beer, 2007). 4 Chapter 2. Previous Work on Morality - An Overview A prominent theory that represent the approach of “Emotionalism” is the “Moral Foundations Theory” developed at first by Haidt and Joseph. Moral Foundations Theory The MFT is a social psychological theory about the intuitive ethics of each human being. It is assumed that there are five inherent centre points, sensibilities or social perception abilities that are responsible for the building of the intuitive ethic. Every centre point is represented by two opposed poles. These five centre points are very similar in different cultures and can also be detected in the behaviour of primates. The facts point native and intuitive centre points of each human being (Haidt & Joseph, 2006). These native points can change during growing up with the experiences someone make (Graham et al., 2013). 1 Definition 2.1 Harm and Care The perception of suffering associated with the motivation to welfare, education and for the protection of others. Thereby the compassion for other people is often accompanied by anger against the originator of suffering. Definition 2.2 Fairness and Cheating The substitution or relationship with others should always be fair, lawful and trustful. Definition 2.3 Loyalty and Betrayal Cohesive coalitions gain advantages in competition. This is a common reason to abide by a group. Definition 2.4 Authority and Subversion Through the hierarchical organization of a group and the establishment of positive relationships to levels above or below, one can benefit from relations. Definition 2.5 Sanctity and Degradation Within this centre point, disgust has a leading role because it is regarded as a behavioural immune system, which protects us from morbid influences (Graham, et al., 2013). Confronted with constant surrounding conditions over a longer period, a domain specific cognitive adaptation occurs (Graham et al., 2013). That is, humans develop mechanisms, in this case moral ones, to solve problems automatically in a similar manner. If a moral decision is to make it happen intuitively. A conscious fray with the own decision will be triggered by explanation, defence or justification opposite to another person (Graham et al., 2013). 2.1.2 Kantanian Rationalism In Kant’s writings “Groundwork of the Metaphysic of Morals” (1785; ger.: Grundlegung zur Metaphysik der Sitten) and “Critique of Practical Reason” (1788; ger.: Kritik der Praktischen Vernunft), he dealt with the topic “moral behaviour” and unlike Hume he present a rationalistic approach. 1 The information in this subsection are based on the Bachelor Thesis of Franz and Götz, 2014 2.1. Theoretical Traditions 5 The rationalism by Kant assumes that making a moral decision is the subject to the supremacy of reason. Moral judgement can be treated as a form of reasoning and is formed by cognitive development during growing up. The emphasized reasoning is the key to moral experience and their articulation means sophisticated moral reasoning. It is assumed that moral reasoning can be observed and recorded by questioning participants hypothetical dilemmata, e.g. “Heinz-Dilemma”. A consistent structure appears through the responses to these hypothetical dilemmata (Monin, Pizarro, and Beer, 2007). Stage Model Plenty of researchers dealt with the topic moral evolution. Jean Piaget (1954) was one of the first scientists who evolved a theory on this field of study. But more popular is Lawrence Kohlberg’s “Stage Model” (ger.: Stufenmodell). This model is partially established on the research studies of Piaget. Within the “Stage Model” there are seven moral stages (in a revised version, he reduced them to six) which humans can reach. During their growth most humans reach the fourth stage. According to Kohlberg, only a few can get access to both other stages. The six stages are grouped by three levels. Stage one and two build the pre-conventional level, stage three and four form the conventional level and stage five and six represent the post-conventional level. Kohlberg defines the term “conventional” as “one conforms to and accept rules, expectations and conventions of a society or authority, because they are the rules, expectations and conventions of the society” (Kohlberg 1995). In addition to the fact that Kohlberg’s theory is distinguished and caused research for the topic “morality”, it received some critics. To mention some critics, the theory is more western-oriented (Becker, 2012) and Gilligan (1984) argues that the one-sided sampling of males provoke a distortion of results. Also Kohlberg changed his theory several times to fit the results of his tests (Garz, 1996). 2 2.1.3 Encounter Moral Dilemma The moral dilemma is one of the most common encounter mentioned in literature and used for psychological studies of morality. It highlights the conflict between moral principles and is applied if the interest is in understanding how individuals engage moral reasoning (Monin, Pizarro, and Beer, 2007). Beside the techniques and direction of arguments above, Borg et al., 2006 analysed reactions on moral dilemmata with brain imaging techniques. They observed that “some moral judgements involve more emotion whereas others involve more reasoning”. The given example is that “neural systems associated with emotions are activated more by personal moral dilemmas than by impersonal moral dilemmas”. Definition 2.7 Personal Moral Dilemma (Borg et al., 2006) A case, where an otherwise desirable action is (1) likely to cause serious 2 The information are based on the Bachelor Thesis of Franz and Götz, 2014 6 Chapter 2. Previous Work on Morality - An Overview bodily harm (2) to a particular person or group of persons (3) not by deflecting an existing threat onto a different party. There are different dilemmata developed for studies in morality, e.g. Heinz-Dilemma or Trolley-Dilemma. All dilemmata have two constants in common that reflect a particular model of morality and influence the conclusions reached by investigators (Monin, Pizarro, and Beer, 2007): 1. The participants have to advise an actor, in a first-person perspective, including the comparison of different outcomes. The focus is localized in the decision-making process. 2. The encounter is build to collide moral duties and to trigger a decision of two morally right actions which are incompatible with each other. Trolley-Dilemma Today, to determine the individual moral characteristics of a test person the Trolley-dilemma is a highly used encounter, especially the dilemma developed by Foot (1967) and the extended version of Thomson (1976). The moral differentiation is limited to the utilitarian and deontological motivation of the test person. Utilitarian actions are based on the idea “Act like the consequences of your actions are optimal for the welfare of all persons concerned” (Hobbes). According to this, every action is calculated by the possible consequences the action might effect. Furthermore, it is tended to an applicable positive usage and a positive goal. The calculation is not confined to the consequences for oneself but also for all other persons concerned Franz and Götz, 2014. The deontological idea complies with whether a moral action is good or bad, independent of the consequences and only bound to rules and laws: “Only act on a maxim as if your actions were to become universal law” (Kant, 1923). Setting: The test person is placed in a situation where a train is impend to run over a group of five people on a rail. On another rail (linked with the first rail) there is one person. The test person can choose to detour the train by switching a lever, to save five people instead of one. So this dilemma wouldn’t fulfil the Definition 2.7 instead it is called an impersonal moral dilemma according to Borg et al., 2006. The modified version by Thomson is different with regard to the way the test person can choose to save the five people. The situation places the test person in a scene where it stands on a bridge with a corpulent man watching a train moving out of control. The train impend to run over a group of five people standing under the bridge, where the test person and the man are standing on. Now the subject can choose to push the corpulent man over the bridge to stop the train and save the five people. This version would correspond to the Definition 2.7 (Borg et al., 2006). Moral Decision: The problem of each Trolley-Dilemma is the same. Is the subject willing to sacrifice one person by switching a rail or more radical, is the person able to cause a death to save more people. The environment varies by the amount of people on each side and the circumstances. The aim is to analyse how and when a test person makes a decision, justifies it and discover the moral boundaries of the test person (). But the difference 2.1. Theoretical Traditions 7 is more subtle than just decide whether it is more personal or impersonal. The categorization personal and impersonal was described as a “first cut”. fMRI Study A deepening analysis - also conducted by Borg et al., 2006 was carried out by the evaluation of three factors of the traditional moral principles: “Consequentialism” and the deontological principles Doctrine of Doing and Allowing (DDA) and Doctrine of Double Effect (DDE). “Consequentialism” is described as “we morally ought to do whatever has the best consequences”. The DDA assumes “that it takes more to justify doing harm than to justify allowing harm”. At least, the DDE “holds that it takes more to justify harms that were intended either as ends or as means than to justify harms that were known but unintended side effects”. Considering the Trolley-Dilemma again this means that in the case of the DDA it is no violation if you “letting someone [e.g. the corpulent man] drown to be able to save five other drowning people” but “Redirecting the trolley onto the sidetrack violates the DDA”. Instead “Pushing the fat man [...] violates the DDE because [... it] intends to use the fat man as a means to stop the trolley” but “deflecting the trolley [...] does not violate the DDE because the victim’s death in that case is only an unintended side effect”. So what about the “Consequentialism” in this case? The difference between “Consequentialism” and both of the deontological principles is, that the latter “claim that factors other than consequences matter, so it is sometimes morally wrong to do what has the best consequences overall” Borg et al., 2006. Moral Reaction Definition 2.7 Moral Reaction It is described as an intuitive behaviour that is caused by witnessing a shocking transgression. Monin et al. mentioned further encounters that are less present in the literature but can lead to a better understanding of morality: moral temptation, moral self-image, lay theories of morality. They may be useful alternatives to create a study but they are not necessary for the current context, though they remain named here. 2.1.4 Moral Emotions A general definition for moral emotions is not existent yet. Most scientists tend to follow the hypothesis that moral emotions are based on moral value judgement which serves the social cohesion of groups (Haidt, 2003; Fiske, 2002; Tangney et al. 2007). According to Weiner (2006), they require a consideration between right and wrong, good and bad, as well as must and should. To restrict moral emotions one might use inductive methods like literature research in psychological and philosophical literature databases. Combined with lists of moral emotions mentioned by Haidt (2003) and Weiner (2006), and the keyword “moral emotion” supplemented by the emotion words mentioned in the “affective lexicon” (Clore, Ortony, Foss, 1987), the following emotions are reported as moral emotions: admiration, adoration, anger, contempt, envy, esteem, gloating, gratitude, guilt, indignation, jealousy, malevolence, pity, pride, respect, shame and veneration. It is obvious that moral emotions form a wide line-up. 8 Chapter 2. Previous Work on Morality - An Overview Determining moral emotions is i.a. possible with the questionnaire “State Shame and Guilt Scale”, the SSGS (Marschall, Sanftner, Tangney, 1994). The emotions shame, guilt and pride are examined by the SSGS. An interesting approach is to use the questionnaire as state-measurement directly after a confrontation with a moral dilemmata Franz and Götz, 2014. 2.2 Morality in Games Before we concentrate on morality in games, I’d like to specify which games are analysed. I will take a look at video games containing animated characters which are mostly humanoid. Although the player has to interact with the characters to complete the game successfully. The moral aspect in games can be presented in different ways. The range of obviousness and complexity varies through the different interaction modes and genres. In the following part we will look at the interaction modes. Furthermore the life experience of the players will be investigated. 2.2.1 Mode of Interaction Single-Player The most common interaction mode in games is the singleplayer mode. This mode can be described as only one player is able to perform in the game environment. It is possible to play against another player or computer animated characters. If the player is confronted with a dilemma he or she has to face it on their own. I assume that the player in general will play a single-player game alone without anyone around. That means the player has to make a decision and process the consequences without a direct interaction with other human players to review the own decisions. Moral aspects, especially dilemmata, become a more favoured feature in games. Popular examples are the “Mass Effect - Trilogy”, “Heavy Rain” and“Spec Ops”. The given examples vary in genre and the amount of moral decisions. Concentrating on single-player games, Svelch, 2014 offers an informative and remarkable analysis on moral engagement in single-player video games. Obviously a player has the power to change the state of a game but to which extend? He argues that the act of making a moral choice in a video game differs from making a choice in the real world. But there exists a “link between someone’s moral sentiment and events in the fictional world”. Assuming that this link is maintained one can speak of moral engagement. Although he suppose that emotions discharging immediately are important for the moral judgement. Such dilemma including the emotional reaction it has generated can initiate or override the rational reasoning. Important to the process of moral judgement and reasoning is the emotional engagement mentioned before. Keeping this in mind it is just natural that emotionally authentic experiences are desired in games in order to cause moral engagement. This will be an important factor to the players decisions within the game, and also to get useful insights in the player behaviour. To expand the complexity and according to the paper of Svelch, 2014 one might notice that the player enters the virtual environment through an avatar. It is possible to play an avatar with a certain, pre-set moral behaviour visible through the dialogue options and action possibilities. Thus the difficulty of behaving 2.2. Morality in Games 9 like the avatar increases drastically with the grade of difference between the own and the avatars behaviour. This dichotomy between player and avatar was particularly important for the investigation by Svelch, 2014. 2.2.2 Aspects of Player Behaviour There are many players of computer games and if one would like to carry out a study, it should be recognized that there is a great variation between them. A suitable paper was presented by Gentile et al., 2009 and I like to sum up the study and present its results. The paper is focused on prosocial video games, that should “increase both short-term and long-term prosocial behavior”. They may defined as games “in which players and game characters help and support each other in nonviolent ways”. Theses prosocial behaviours are defined on various levels. In addition they assume that “prosocial and aggressive constructs [...] also do not tend to be entirely independent”. This is because people can be both “high in prosocial and aggressive behaviors”. A fact that might be forced through short-term context of a game. Therefore they refer to the General Learning Model which suggests that the learning encounter may have “affective, arousal, and/or cognitive affect”. Although games offer opportunities for reinforcement as well as punishment and this might influence the „cognitions, feelings, and physiological arousal “which are able to interact. For example, if one is confronted with a decision to made within a game situation, the consequence of the decision can have an effect to “future cognitions, feelings, and arousal”. The learning process within the prosocial games is particularly increased if a game “requires prosocial behaviors to succeed the game”. Therefore if such games are played several times with short-term effects they may have an effect also to long-term effects. This is meant in the sense, that they cause changes to cognitive constructs (e.g. perception and expectation schemata), cognitive-emotional constructs (e.g. stereotypes) and affective traits like conditioned emotional responses. It is assumed that there is “greater learning and transfer” if players play more than one game but the solutions are the same. They go one step further and suppose that the changes on the long-term effects also lead to a change within the personality of a player. Their study contains three different cultures and age groups as well as three different study designs to test the hypothesis “prosocial video games increases prosocial behaviors”. Only the hypothesis of the third study includes the aggressive and neutral video games to study increasing aggressive behaviour, too. The results of the study were unexpected to some extend. This is because they observed quite different behaviour in prosocial and aggressive games which end up in the speculation that violent games are not related to emotional awareness but to empathy. What is influenced by the fact that most violent games need more cooperation than others Gentile et al., 2009. They point out that there has to be further research to prove their speculations. The degree of utility might not be questioned here but I think it could be more insightful to use each study for only one culture and different age groups. This is because moral behaviour including emotions can be quite different depending on the culture where humans grow up as I mentioned before. 10 Chapter 2. Previous Work on Morality - An Overview Players Behaviour by Age An interesting aspect for research may be the players life experience. Assuming that a player in advanced age has generally more life experience than a younger person, it has an influence on the behaviour in playing a game with focus on moral decisions. Beside the players age there are other aspects that might be fundamental for the research in moral behaviour, e.g. gender and cultural background. Both aspects can give other insights and in combination with the age a detailed survey of moral behaviour. Children A study among children which have to choose an advisor for a dilemma of moral or scientific nature, prove to be an insightful work for this topic. The study was carried out by Danovitch and Keil, 2007 with 48 children for each of their three experiments. The children are aged between five and nine years and their cultural background was predominantly European and American. A part of them was in the kindergarten and the other part in the elementary school. The information given before the description of each experiment are helpful to understand the format of the experiments, so I would like to sum them up. Basics Judith Danovitch and Frank Keil are introducing the cognitive evolvement of children starting with the age of 4. At this age most children show a “crude understanding of the division of cognitive labour”. Just at the “end of the elementary school [their] intuitions about the discipline based ways of clustering knowledge become quite strong”. It is assumed, that they “sense how to cluster knowledge domains by thinking about distinctive causal patterns that give rise to different sets of phenomena”. Children seem to assume that people who understand phenomena are also able to understand other phenomena of the same causal forces. But not all arise in this way, because some domains are not so closely linked (e.g. moral and social conventions). Therefore children are dependent on other people as a source of information. They “understand that other people have knowledge and beliefs that differ from their own”. But children often have to choose between more potential sources. The selection among them might be strategic. They know how to avoid information from informants who are obviously ignorant. One thing is still unclear. That is, how they will select between more advisors which have the same potentially knowledge status Danovitch and Keil, 2007. Experiments All three experiments are based on choosing between moral advisors to help for a dilemma of moral or scientific nature. But the experiments address different interests in behaviour. The first and second experiment were arranged so that the children can select between an “extremely nice person and extremely knowledgeable person”. The children were asked to select the best advisor for the moral dilemma or the dilemma with scientific nature. The third experiment were built to investigate specific moral characteristics required to solve moral or scientific problems Danovitch and Keil, 2007. 2.2. Morality in Games 11 Results The children of the 4th grade were able to easily “distinguish between moral and scientific advisors”. The distinction was quite “consistent and accurate”. In contrast the kindergarten children were successful in “distinguishing among experts when dealing with familiar domains”. But of course it was difficult for them “clustering scientific knowledge”. A real obstacle for the ability to distinguish is that the view of moral characteristics is essential for solving both types of problems (especially for younger children) Danovitch and Keil, 2007. Instead, the older children “appreciate that it is not necessary to behave morally or act in accordance with moral standards in order to solve scientific problems”. This might be the result from perception of moral behaviours as indicator of willingness. On the other hand it could result from personal experiences where knowing moral response is emphasized above other types of knowledge. It is still “difficult for them to make fine tuned distinctions between moral knowledge and scientific knowledge”. This could possibly result in the belief that moral behaviour make someone more adept to solving other problems Danovitch and Keil, 2007. Another interesting result was the decision among available advisors. The “younger children favored the nice advisor and choose moral characteristics for both types of dilemmas”. There was an overestimation of their own knowledge and that of other people. If they seem to be unable to discern causal patterns, the children “rely on other indicators of moral knowledge such as the presence of emotion on the persons proficiency with factual information”. It is clear that there is a “shift [within the children] during elementary school”. They learn to construe morality as a distinct area of knowledge. Therefore the understanding influences their selection of source of advice. They become able to “see the difference between decisions based on analytic thought and set of moral intuitions that may well involve assumptions about peoples hearts and minds”. But it remains a limited understanding of what makes someone a good or bad advisor when dealing with challenging moral and non-moral problems Danovitch and Keil, 2007. From Middle School to College Students The studies of Douglas Gentile et al. were focused on students at the age between 10 and 19. In addition, the chosen students had different cultural backgrounds: Japanese, Singaporean and American. The number of students that have participated was overall 2718, spread over three different studies. It should be indicated that the participants were almost equally distributed on gender. Study 1 The first study has 727 participants with an average age of 13. They were partly elementary and middle-school students from Singapore. The underlying hypothesis of the first study was, that there is a “positive association between the amount of time spent playing prosocial video games and the amount of prosocial behavior even controlling for other factors related to prosocial behavior”. As predicted, “prosocial game exposure was positively related to prosocial behaviours and traits”. Instead, violent game play was “negatively related to the prosocial behaviours and traits”. Therefore the prosocial and violent game exposure were recognized as “highly correlated”. The results of this study were consistent with the hypothesis 12 Chapter 2. Previous Work on Morality - An Overview that exposure to prosocial video games increases prosocial behavior. “A causal interpretation based solely on these results would be inappropriate because either variable might precede the other” Gentile et al., 2009. Study 2 In this study, divided into two parts, 1830 students of middleschool and undergraduates have participated. These students from Japan were at the age between 11 and 17. This study was based on the hypothesis that the “habitual playing of prosocial video games would increase prosocial behavior assessed months later”. The amount of pro-social game playing (group 1) and the prosocial (group 2) was measured at two time stages, where the first were the start and the second were the end. They discovered that the “causal path was significant” and that there must be a “bidirectional relationship between prosocial gaming and prosocial behavior” . Gentile et al. conclude that there is “evidence for a causal longterm relation between prosocial game play and prosocial behavior” Gentile et al., 2009. Study 3 At least, the third study was carried out with 161 participants at the age of 19 from the US. They investigate the hypothesis that “playing prosocial video games would increase prosocial behaviour in the immediate situation, whereas playing matched violent games would increase aggressive behavior”. It was carried out among six games of different nature, from prosocial to neutral and violent. Determined results were split into a main analysis and an alternative analysis. Within the main analysis there is a tendency that “playing a pro-social game leads to more helpful behaviour” in contrast to playing a violent game, where “players tend to be more hurtful”. They measured that “violent gamers hurt their partners significantly more than did either those who had played a prosocial game, or those who had played a neutral game” . The alternative analysis was more focused on the role of the content of the game and its effects. Thoughts on the Studies The first study by Danovitch and Keil, 2007 is not directly linked with the theme moral dilemmata in games. But it should give an impression of the moral behaviour i.a. evolution of children. The moral evolution of children may be revealing for the future moral behaviour of children. I assume that a moral study like these of Gentile et al., 2009 are not suitable for children because of the age restriction (e.g. FSK). However, it would be interesting to study the moral evolution over years and maybe there will be evidence for predicting the later moral behaviour by considering the moral evolution in the past. The studies of Gentile et al., 2009 provide some interesting approaches and methods which should be used for further research. But I would change the setting of the combination of test persons. I assume that there would be some exciting insights on the behaviour of players, if one chose to carry out a study with one of these methods and different age groups of one nationality. It could lead to a further understanding of the influence of the life experience on moral behaviour. 13 Chapter 3 Moral Dilemmata and Conceptual Game Design 3.1 Theoretical Model Types of Dilemmata On the basis of the MFT described in 2.1.1, E. Franz and V. Götz developed five scenes according to the dilemmata types mentioned in the Definitions 2.1 - 2.5. They are the foundation of the game to build and for each dilemma type there is a separate scene. In the following paragraphs I’d like to describe the development process and the scene types. Development Process Crucial for the creation of the five dilemmata was the aim, selecting a preferably realistic scenery. The test person shall have the impression that a given situation could exist in a real environment. The reason for this is that a test person may behave in a given dilemma in the game like he or she would do in the real life. The developed situations were investigated in a preliminary study with 30 test persons. Presenting the dilemmata in a written form, without pregiven decision options, the test persons were supposed to decide whether a dilemma is realistic or not and they were invited to suggest possible reactions for each dilemma. The results, especially the reactions to the dilemmata of each test person were analysed and build the options in the game. The chosen framework story is a date situation because it seems to be an understandable starting point for the whole story and each test person. Proceeding from here the test person will be confronted with five different situations as he or she is on the way to the date. Now, we like to present the five decision trees for each dilemma the test person can go through. Dilemmata Dilemma 1 - Harm & Care The test person encounters a group of children. If he or she get closer to the scene it will be clear that one child in the centre of the group seems to be the victim here. Optionally the test person can go on without an interaction. However he or she has the chance to help the child in the centre: 14 Chapter 3. Moral Dilemmata and Conceptual Game Design F IGURE 3.1: Dialogue Dilemma 1 Dilemma 2 - Loyalty & Betrayal The next situation presents a group of friends waiting in front of a strip club. If the test person goes on it will be obvious that a bachelor party was planed today. As soon as the test person interacts with one of his or her friends he can choose between the following options: F IGURE 3.2: Dialogue Dilemma 2 Dilemma 3 - Fairness & Cheating Continuing, the test person walks onto his or her favourite restaurant. Getting closer to the restaurant it will be clear that a waiter and a guy in front of the entrance door have a dispute. Hearing some arguments of the waiter it is obvious that he has mentioned racist reasons and refuse admission to the guy. The test person can choose between the following options: Dilemma 4 - Sanctity & Degradation Walking on a busy street the test person can recognize a crowd of people. If he or she comes closer to the scene, the test person can discover a women. She seems to be wobbly on her legs and the reason for the crowd. A dubious man comes closer to her and the test person can imagine that he has no good intentions. The following options can be selected: Dilemma 5 - Authority & Subversion Suddenly the test person is confronted with his or her boss. The boss will inform the test person that an error was discovered in an important project at work. The project has to 15 3.2. Building a Game F IGURE 3.3: Dialogue Dilemma 3 F IGURE 3.4: Dialogue Dilemma 4 be presented to the customer on the very next day. There is a chance that the company could lose the customer if the project is not corrected immediately. The boss will order the test person to help and he or she can make the following decisions: F IGURE 3.5: Dialogue Dilemma 5 3.2 Building a Game The next phase of the process after creating the background story and the essential dialogue options for the dilemmata, was to build the game step by step. 16 3.2.1 Chapter 3. Moral Dilemmata and Conceptual Game Design Environment and Interface Before one evolves the environment and interface as basic implementation, it should be analysed which kind of environment and which interface functions are to realize. Environment In general we can restrict the possible environments to suit the story well. Some aspects has to be considered. The first will be the genre of the future game. In principle, every genre is possible but some are more suitable than others regarding to moral dilemmata. Appropriate genres for our story are Role Play Game (RPG) or Simulation. Role Play Game An RPG is characterized by playing one role or more. This role can take the form of a human being, fictional character, an animal or even an object is possible. The opportunities are nearly infinite. This character has a predefined personal profile but it is possible to adjust some aspects and define the appearance before starting the game. It is placed in a world where it takes a certain role within the story of the game. Controlling the character can be realized in ego- or third-person perspective. Most RPG games define various abilities or values a player can modify. The exploration of the world and continuing the story increase the characters experience (e.g. in form of points) which is used to define the abilities and values mentioned before. Another important aspect is, that the avatar can be equipped by different objects. These objects can change values, abilities and also the look of the avatar. Moreover, abilities and values as well as the equipment are decisive for the successful completion of the game. Simulation If one searches for a generally accepted definition for simulation one probably will not find it. occasionally there are definitions within papers or books but not all authors define the term although they frequently use it. An interesting approach is submitted by Prensky, 2001. He argues that “simulations are not, in and of themselves, games” because the common elements amount to games like “fun, play, rules, a goal, winning, competition etc.” are normally not part of a simulation in general. This does not mean that simulation games are not existent. There are some aspects which prove this like “if a game ’simulates’ something, it is a simulation as well”. Simulations are suitable tools to test hypotheses in a safe environment, e.g. the virtual reality. Such simulations are not identical because for most sciences there exist individual simulations which can be more technical or social etc. Strictly speaking the program I wanted to build is a social simulation. The simulation amounts to the following characteristics. The background story and the dilemmata including their dialogue options were written and predefined before. According to Herz and Blätte, 2000 our game will be the simulation model for the experimental model, a simulation experiment. The replicated parts of the game are the story and dilemmata including their mentioned characters. The algorithm for the dialogue in each scene (based on the predefined dialogue options) defines the rules for the substeps leading to the desired results - discover the moral behaviour of a player. Afterwards the results will be analysed by Franz and Götz. 3.2. Building a Game 17 Time This variable is crucial for the appearance of the environment and also the characters. Time is meant to be the daytime and an approximate year. Daytime can be dynamic or static depending on the meaningful and/or wanted appearance of the fictional world around. It can have a significant influence on the scenes authenticity and the players perception. The year varies according to the genre of game. Especially in the fantasy genre the definition and name of “year” is very differently. Our simulation is placed in a year within the last few years. Therefore the objects in the environment are dominated by modern styles. Corresponding to the background story the daytime is static and is situated in the late afternoon. Focus The focus defines whether objects, characters or other parts of the virtual environment are detailed. The reason for this is to direct the attention to the important features of the virtual world. For our game it is appropriate to detail the characters involved than the surrounding environment. Interface Also the interface varies among the different genres but some aspects are very welcome. Notable characteristics are clarity, simplicity and consistency. This is because the player should not be distracted or have a restricted view to the game. Also the supported functions accessible through the interface should be consistent and self explaining to avoid false expectations with respect to the functionality. And at least it is important to provide all necessary functions for the respective scenes. E.g. the start screen shows (mostly) different options to a player than the ingame screens because the functions are not necessary to successfully play the game. Not required options should be hidden but easy to reach. In addition make sure to define which elements of your interface are interactable or not and which are static or animated. Not all options need to be intractable, e.g. a small map for the overall view of the current environment do not necessarily be interactable. It is sufficient to provide customization options and/or shortcuts for the small map (also known as Minimap). Static elements are prominent because animated ones are limited useful. Even these characteristics depend on the game genre. Static means that the elements remain on their defined positions. Animated elements could be a Minimap or a life display. Lastly, useful feedback should be offered if a player interacts with some interface elements or wants to do impossible actions like running out of the explorable world. Within our simulation there is a selection of interface elements indispensable. Most of them are static though the Minimap will be animated. Immediate feedback will be provided if a player leaves a specific zone. 3.2.2 Character Features Facial Expressions The face of humans provide detailed information about emotions and moods. Animation of facial expressions become relevant if the game is played with 18 Chapter 3. Moral Dilemmata and Conceptual Game Design ego-perspective and the number of situations where the player stands faceto-face with another character. Important regions are the eyes, forehead and the mouths including the cheeks. There are various methods to animate a face, e.g. parameter-controlled, muscle-controlled, spring model (“Federmodell”) and visual speech (Jackèl, Neunreither, and Wagner, 2006). Gesture & Full Body Movement Gestures are movements of the interpersonal communication. Especially the movement of arms, hands and the head are evidence for this form of communication. In addition, the animation of the whole body reflects an extensive visual action of a character. There are some important design aspects to consider. Precisely, if you want to animate a human being in a game, you should pay major attention on the movement, which should be as natural as possible. This increases the authenticity of a character, especially in games with ego-perspective where humans play an important role. Methods There are different methods to animate a character but a suitable program is required to form animations. In principle watching the movement in reality or in a video are helpful opportunities to study the motion of humans or animals. For detailed knowledge about the motion of each living being one can study the literature about anatomy of the respective being. Usually there are information about the musculoskeletal system (muscles, bones, joints etc.). There is plenty of literature broaching the issue of anatomy and equally videos containing movement of different kinds of living beings. Motion Capture To realize the natural movement of humans or animals is quite difficult. The reason is the complexity of motion tasks and the obviousness of the smallest aberration within movements. In favour the Motion Capturing methods were developed. Motion Capturing is defined as the record of motion for direct or future analysis or for playback. The method behind is to add sensors or marker on human actors, whose position or orientation can be traced. This data are then transferred to a computer model. Digital Puppeteering is a technique where the data is directly translated to a digital figure. Here, the human actor is able to control the figure in realtime. If the modelled figure is humanoid the actor represents the whole figure otherwise the regulation is conducted by auxiliary tools. Another technique might be Channel Animation. It is a simpler version of Motion Capturing or Digital Puppeteering. This technique is characterized by degrees of freedom for a figure which are controlled through input units (e.g. Joystick). Channel Animation is mostly an additional technique for other Motion Capturing techniques because it is useful for the animation of aspects a player can not adopt (e.g. eyelids or other body parts of fantasy characters). The records of motion are generally not used one-to-one. This is because the human actor doesn’t match the shape of a figure. Instead the data are reworked before use (Jackèl, Neunreither, and Wagner, 2006). Animation Program Programs that might be useful for animation are numerous. They can be subdivided by costs and the amount of functionality. 3.2. Building a Game 19 Some programs might be Xsens MVN (for Motion Capture), Blender, Unity, Renderman (from PIXAR), Maya or 3D Studio Max. Blender is the only program that is free and has a great amount of functionality which can be used widely. From character creation and animation to developing whole games and animation films. Therefore it is predestined to use it for private purposes. Most of the other tools allow the free use for students and educators or give a free trial version. It is clear that to make a good designed and animated object, especially living characters like animals or humans, you have to spend much time exploring the possibilities and practising these. 3.2.3 Character Interaction The interaction between a human player and a virtual character is very important if a game is mainly based on it, e.g. in RPGs. Depending on the form of interaction you want to create (e.g. fighting or talking) some aspects have to be recognized. In our simulation the focus is on the dialogues between player and characters. The precondition for a dialogue is that a specific character can be chosen or the character “talks to us”. The form of talking to a character is mostly realized as a trigger, where moving or clicking on the trigger initiates the dialogue process. Or else, we walk into a zone where a trigger was attached to and a dialogue starts immediately. The dialogue system we wanted to create was based on interactable characters. Once you have clicked on the character the dialogue will start. The dialogue options were arranged in small decision trees. These trees appear in different shapes. In dilemma one, three and four, the path you will follow in the decision tree is dependent on the person you have clicked on. In contrast, in dilemma two and five the path you will follow is dependent on the first option you choose. Both of these variants might be common ways to construct a decision tree in a game. 21 Chapter 4 Realisation of ‚Projekt: Rose‘ 4.1 Technical Details The game was developed on a laptop with the following status: Operation System: Windows 8.1 Processor: Intel Core i7-3610QM CPU @ 2,30 GHz RAM: 8,00 GB System Type: 64-Bit-Operation System, x64-based Processor Graphic Board: NVIDIA GeForce GTX 660M MakeHuman The version of MakeHuman I used was the 1.0.2. This tool offers a great amount of settings and options to create a human character, even in the basic freeware version. Blender The version I used was the 2.74 and it is a freeware. Blender is a complex tool to animate and create characters in detail. It can also be used to create animated films or whole games. I used it to create clothes and hair as well as the animation of characters. Unity Engine The Unity Engine I used has the version Unity 5.3.3f1, where it was updated several times during the development of the simulation. It is free until you earn more than 1.000 $ with a program produced with Unity. I used it to create the whole simulation, especially the interface design, the scene construction and to finish the animation of characters. There are plenty of assets you can download and integrate in your environment, some are free and some are chargeable. I also downloaded assets to create the environment of the scenes. All assets downloaded are mentioned in the A. MonoDevelop and Xamarin MonoDevelop was some time the main tool to create the scripts for the behaviour of the simulation but after an update it changed to Xamarin (actual version: 5.10.3), which I used until the end of developing the simulation. Both tools support i.a. the programming language C# therefore I used it, because it has many grammatical characteristics in common with Java, which I have learned during my studies. 22 Chapter 4. Realisation of ‚Projekt: Rose‘ 4.2 Character Creation 4.2.1 Human Body 1 The body of all characters were created with MakeHuman because it provides many functions to form and customize a character. The basic program is freeware and contains all basic functions needed to create a human being. For every character that is visible in the scenes I used this program. There are various functions to model your character, add geometries and attach materials to eyes, clothes etc. Also you can pose and animate characters but in the basic version this function is very limited. I only used the general model function as well as some facial modelling to customize the characters. For further processing them in blender you need to pay attention on the export function in MakeHuman. There is an extra mesh-format “Blender exchange (.mhx)” on the left that you should select in order to use it in blender. In general the options on the right can be used by personal desires. My selected options were “Füße auf Boden” and “Meter”. F IGURE 4.1: MakeHuman User Interface 4.2.2 Clothes and Hair Creation Before creating the clothes and animations of the human body built with MakeHuman make sure that the models will be exported correctly. Some other settings have to be chosen in Blender before importing these models. Open Blender and go to File > User Preferences and prove if “Import-Export: Import: MakeHuman (.mhx)” and “Render: Cycles Render Engine” are selected. If not than one should select them or install them. They will be necessary for the clothes and animation methods I have applied 2 . Then one should be able to import the .mhx file of a created character and switch the render mode from default Blender Render to the Cycles Render. To ease the work with the imported model, mark the model and chose the preferred 1 Please note, that all italic words or letters in the following sections are functions or buttons of the particular program. 2 Some publisher of instruction videos recommend them 4.2. Character Creation 23 unit to scale it down with Properties > Scene > Units. I have chosen the metric system because it is familiar to me. Clothes Clothes To create the clothes I used the available mesh 3 of the imported character. At first switch from Object Mode to Edit Mode and mark the mesh tiles which the clothes will cover at the end. Then duplicate the mesh (or use it instantly) and make them a few times larger, so that it stands out a little. This method should ensure that the clothes look like the character wears them and not brushed on the model. In the next step I used the Sculpt Mode to edit existing irregularities and unwanted surfaces of the current mesh. This should be done because by duplicating the character mesh all features of the body will also be duplicated, like the deepening by the navel. This procedure can be done for every other clothing. Shoes For the shoes I have decided to apply another way of shaping. Create a cube and scale it down until it has about the size of a foot from the ankle to the bottom and from the anklebone to the Achilles’ tendon. Switch to “Edit Mode” and chose the “Face Select”. Then I used the “Object Modifier” option with the modifier “Subdivision Surface” in Properties. Now I styled the shape by selecting and modifying various mesh tiles and if the first design was acceptable I applied the “Subdivision Surface”. To refine the surface of the shoe I repeated the addition of the “Subdivision Surface” and if necessary also repeated the design of some mesh tiles. Finished the shoe I used to go to the outliner and drag the shoe and drop it into the character. By dropping it into the character I selected the “Armature Deform” and “With Automatic Weights”. This option should ensure that the shoe is moving together with the character. Hair One of the difficult procedures was creating the hair. At first I selected the mesh tiles of the scalp in the “Edit Mode” where the hair should be. The selected tiles built a new vertex group in the Properties > Object Data. Than moved to Particles and chose the type “Hair ”. There I changed the number and hair length in Emission. In Render I selected the b-spline and change the steps from default three to six and do the same for steps in Display. In Children I selected Interpolated and change the display and the render to 20 as well as chose the Kink adjusted to the type of character I wanted to create. Within the Vertex Groups > Density I chose the predefined scalp, so that Blender knows where the hair should be placed. After creating the basic hair change the mode to Particle Edit. There I brought the hair into shape by using the comb tool. Then I switch back to Object Mode and in Properties > Object Modifier and press “convert” in the particle system modifier. I converted the mesh type to Curve and adjust the geometry in Properties > Object Data until the form is acceptable. Afterwards 3 Each character has a mesh that is consistent with the shape of the character. The mesh can be deformed if necessary by using methods from the Edit Mode or the Sculpt Mode. 24 Chapter 4. Realisation of ‚Projekt: Rose‘ F IGURE 4.2: Parameters for adjusting the hair. converted the curve back to Mesh. Beyond finishing the hair I used the same drag and drop process like after I finished the creation of the shoes 4 . 4.3 Character Animation It is advisable to use Blender for the character animation, too because it was also the tool to complete the character design. In the header of the program I chose the Animation screen layout. First, I opened the right bar in the 3D viewport by selecting the plus sign in the upper right corner. Then I selected the IK (inverse kinematics) for the legs within the MHX FK/IK Switch folder. This is used because if you change the position of e.g. the hips up or down, the legs will also move. Otherwise you would only change the position of the whole character. The next step will be changing the mode to Pose Mode where you can change the position of parts of the body by turning the parts along the x, y or z axis. In the lower window Timeline switch the active keying set to LocRot (Location Rotation) and change the final frame to 31. Afterwards I switched the editing context from Dope Sheet to Action Editor in the upper left window Dope Sheet. Then I selected New in the same bar to create a new animation. Depending on the animation I wanted to change the pose of the character to the first pose, the character should start the movement with. Then select the whole character and copy it, and paste it to the upper left window. Now the first pose with all keyframes is stored to the first frame. After 4 Note that this procedure for the hair was necessary for me to make sure that the hair will be displayed in the Unity3D environment. It is not possible to simply use the hair created with the particle system. There are other methods usable if one searches for them, e.g. YouTube offers a lot of videos for alternative hair creation without using particle systems. 4.3. Character Animation 25 this step I changed the current frame from one to eight. Here I changed the pose of the character to the next position. With the next position I mean the change within the movement sequence. You don’t need to pose the character frame by frame because the tools calculate the positions in between automatically. I recommend to do small steps from pose to pose to get a smooth transition between them and the final animation. F IGURE 4.3: Keyframes within the animation screen layout. The process of creating poses and adding keyframes to the upper left window was always the same. In the upper left window, I named the action and saved it by selecting the F next to it. So now this action can be used by all other characters created, assumed they have the same settings. You just need to access them via File > Append and choose the folder with the animated character. Click on the character and choose the action folder. There you can find the finished animation. By selecting the action you drop it to the current characters action folder and apply it automatically. In the final phase of the character animation process I turned the character into the so-called T-Pose and saved it. The T-Pose is recommended to import the character into the Unity environment so that you wont get any complications with the movement. Not all characters within the simulation scenes are animated. A few of them only have a pose like a paused gesture. Creating a pose is more simple than the whole animation but with it’s “frozen” gesture, the character also got meaningfulness. The pose is also made within the Animation screen layout. The settings in the central window have to be the same as at the time you start animating a character. But afterwards you just need to get the character into the right position and save it. 26 Chapter 4. Realisation of ‚Projekt: Rose‘ F IGURE 4.4: T-Pose 4.4 Scenes All scenes were made with the Unity Editor but they share the same environment. Only the characters within the environment are different among the scenes. 5 Environment To build the environment I used different free Assets downloaded from the Asset Store in Unity and other websites 6 . For the downloaded Assets see A. These Assets were dropped in the Asset folder of Unity to be available for creating the scenery. The first step was to build the ground of the scene by adding a plane in a new scene. The plane was designed by adding texture to the plane, in this case the basic ground should look like grass. Afterwards I have used objects that can be combined to build a street system so that I can create the streets surrounding different parts of the future city. The next step was to use different building objects to shape the city with a modern and an old town part. Some areas were avoid to be the park and other green space of the city. For the footpath between buildings I used a brick texture and draw it onto the ground where it was appropriate. Other ambient objects that were added to the scenery are trees for the park and next to some sidewalk, flowers on a footpath as well as park stuff like a playground and park benches. Also a sign for the restaurant and the strip club were added to highlight them. According to the defined daytime in 3.2.1 a Skybox and a sun with a low position was inserted to shape the ambient of the scenes. For all the objects added to the scene and which are standing on the ground it is necessary to attach a Collider. This is important to avoid that a proband will move through the buildings etc. as well as move across the edge of the scene. Therefore I used cubes where the surfaces render was turned off and only the Collider persists. The same applies for the explicit 5 6 To get a first impression of the different scenes, see C. The other websites that should be mentioned here are: ShareCG, archive3d, artist-3d 27 4.5. Interfaces scene border. It should work as a safeguarding if the scene border will be crossed accidentally. Afterwards I duplicated the created scene until I got five scenes with the same environment. Each one for each scene to build and named from one to five. That was the basis for the addition of the different characters. First Person Controller The first person controller (FPC) is part of the Standard Assets if you create a scene in Unity. I used it for the first person perspective for the simulation. The first person camera is pre-attached. The FPC was moved on a different place in each scene according to an appropriate starting point. I attached some canvas to the FPC to create a first ingame user interface including the Menu (where a button is attached) and the HealthBar (including an image). The position of them is fixed so that they wont change during the game because it is not necessary. Also the camera for the Minimap was placed into the FPC so that it fulfils its requirement. Characters The non-playable characters, also known as NPC, were placed into the scenes as last step to complete them. They were arranged in order to build a plausible scene according to the dilemmata developed by E. Franz and V. Götz. Each NPC has some textures in their Asset folder. These textures were attached to the skin and the eyes and come from the MakeHuman tool. The clothes as well as the hair were coloured by adding textures of downloaded assets to them. For each animated character in the scenes (one per scene) it is necessary to attach an Animator Controller. Animator Controller This controller defines the succession and duration of animations made by a character. It is set up within an extra Animator controller window within Unity. There you can change the settings of each animation you have add and for each link between two animations. Therefore you can create extensive animations, e.g. with a tree structure and different trigger values to follow nodes. F IGURE 4.5: Example for Animator Controller 4.5 Interfaces Start Interface The interface description in this part only contains the visible elements. Functionalities are explained in 4.2. The Start Interface contains the basic elements necessary for the simulation to play and for the later analysis. There is a start and an exit button as well as an input interface with the place holder Enter your number, please. The elements without any functionality are the background picture and the title of the simulation Projekt: Rose. In the Unity Editor the start interface is a separate scene 28 Chapter 4. Realisation of ‚Projekt: Rose‘ with only one canvas and some additional undefined objects with scripts attached to them. These are important so that some scripts can satisfy their task, e.g. the scene storage. F IGURE 4.6: Start Interface Ingame Interface This interface contains four elements, three of them are displayed constantly in a scene. The menu with the option “Beenden”, the life display and the Minimap. Only the dialogue options are shown if a test person click on a character in the game. The rest of the time it is hidden. Another message will be displayed if you leave a scene without interacting with any character. F IGURE 4.7: Ingame Screenshot 4.6. Functions 29 4.6 Functions 4.6.1 Interface Functions Start Interface The start interface contains three functions. The first one is the input field. It is used to get the number entered by the test person. If a test person has entered a number to the field, it will be closed for input but you can still see the number. The number will be stored in an integer value in a script to use it for another function described later. The next function is included in the start button. It is self explaining that if you click on the button the simulation will be started. The first scene to load is the value stored in the second column of the CSV file described in the Randomized Scenes function. The last function is the exit button. If you click on this button a message will show up and you will be asked if you really want to exit the game. Depending on the given answer you clicked (yes or no) the game will be closed or turn back to the start interface. The message will be hidden afterwards. It is a common method in most programs to make sure that an user has pressed the right button. Ingame Interface The interface displayed in each scene contains four functions according to the interface elements mentioned in 4.2 Interfaces. The menu with the “Beenden”-Button gives a test person the opportunity to leave the game. If you click on the button a message will show up and ask you if you really want to leave the game. Equal to your choice if you pressed “Ja” the start interface is displayed and if you pressed “Nein” you will stay in the simulation. The life display in the centre of the lower part of the interface is the indicator to give feedback to the state of your bouquet of roses. If you load a scene the display is updated to show the consequences of your choices. A Minimap is a common element in games and it displays the environment from above. In a scene in the simulation it displays an area of the environment with the test person as the centre. It also contains a few marker that represent the characters you can click on. A main person is characterized by a blue marker and an aggressor has a red marker. It is realized as a script attached to a second camera that focuses the playable character. There is an update function in the script that adjust the Minimap while the character is walking or looking around. The marker is a 3D globe attached to a interactable character in the scene. To make them invisible in the environment during the simulation is active, they were moved to a different layer that is not rendered by the first-person camera but by the second camera. 4.6.2 Randomized Scenes The randomization was an essential requirement for the simulation. It is realized as a list stored in a CSV file. The first column of the list contains a number from 1 to 120 according to the number of test persons that will participate in the upcoming study of E. Franz and V. Götz. The five following columns contain numbers from one to five where each number represents a scene in the simulation. Every of the five numbers is contained only once. 30 Chapter 4. Realisation of ‚Projekt: Rose‘ Consequently each possible sequence of scenes is covered. A script will load the corresponding number sequence of the CSV file depending on the input value of a test person entered in the start interfaces input field. This sequence is stored in an array from the start until the end of the simulation. Furthermore it is used for the Data Collection function described below. A real randomization was realized at first but consciously given up mainly because of possible double sequences. Instead each test person will get a number to enter and the script loads a sequence. Therefore it is rather a pseudo-randomization than a real randomization. 4.6.3 Character Detection The detection of a character in the scene is the basis for the Dialogue System. This function will be performed if the test person has start the simulation. The detection function recognizes every object in the environment where the cursor is pointed to. At first, it is not necessary to click on an object but it is used within an OnClick method to identify a character placed in the environment. To detect an object in the environment of any program realized in Unity3D it is important to use the Raycast function. According to the Unity Scripting API the raycast is described as “Casts a ray, from point origin, in direction direction, of length maxDistance, against all colliders in the scene”. The script in the simulation is called RaycastProve. It contains the Update() method that will be called every frame. The method to be called within Update() is the RaycastIdentifier. Basically this method returns the object which was hit by the mouse position. In addition for the simulation it contains a method to get the tag 7 of an object. The returned tag will be used for the Dialogue System. The update method also includes a boolean value to stop the raycast when it is not needed, by default it is false and by starting the first scene it will be true. 4.6.4 Dialogue System The dialogue system for the simulation contains four scripts as components. The DialogOptionen, DialogManager, a separate script for each scene to handle the process of a dialogue and at least the BlackScreen. Also a textfile for each scene. Text-Files Each text-file includes the options that were determined by the preliminary study mentioned in 3.1. They are arranged in a chronological order and sorted by the character where the options will show up. All options for the main person(s) are written first. “DialogOptionen” Script The first time this script will be used is when a scene is loaded. It contains a method with a streamreader to get all the strings of the text-file. Every string in a text line will be placed in a string array that stores the options. To make sure that the string array is empty 7 A tag in Unity can be set if you need it. By default every created object does not contain a specific tag. The Unity API describes it like “A tag can be used to identify a game object. Tags must be declared in the Tags and Layers manager before using them.” 4.6. Functions 31 when the next scene is loaded the array will be reset. It is an easier way for all other scripts based on this script to operate correctly. “DialogManager” Script A click on an object the script is attached to will call this script. If you click on a character the first thing to happen is that the time will stop. That means you can not move your character any more and have to follow the dialogue instead. It should avoid running away from the scene for other reasons and concentrate on the dialogue and it is also a better way to operate more realistic. That is because if you start to talk to an unfamiliar person in a real situation it is not a common behaviour to begin to speak but then simply walk away without continuing the conversation to the end. So, once you clicked on a character the DialogManager will load the dialogue script for the active scene. It also enables the GUI for the dialogue option buttons so that they can be drawn in the interface. This is a necessary method in Unity3D because if you do not call the methods based on the GUI within a OnGUI method the engine will not execute any of these methods. Furthermore the raycast will be turned off because it is not important for the further behaviour, yet. 4.6.5 Speech and Thought Bubbles The bubble images for speech and thoughts are a static element within the interface. For each scene there is one bubble that includes some hints for what happens in the scene. If the own, controlled character thinks about something, than the font colour is grey. But if there is a conversation or just snatches of a conversation, than the font colour is black. The speech respectively thought bubble is loaded every time a new scene is loaded. Therefore it is readable right from the start, so the player has the chance to get a very small introduction for the scene. Dialogue Process Scripts These scripts will be called if the test person has clicked on a character in the scene. The process of each script is similar so the following description will be general. The first method to call is starting the dialogue evoked by the DialogManager. The RaycastProve method will evaluate the tag of the character you clicked on and sets the two boolean values for the main person and the aggressor. According to the tag the method will set one to true and one to false. It depends on the boolean values which dialogue sequence will be shown because each sequence has its own process. The structure of the process was realized by sorting the dialogue options to layers. The layer zero contains all dialogue options that show up when the test person clicked on a person. The layer one is called if you click on an option and there are further options based on it. All options of the layer zero will be hidden then. At least if there is an additional option after clicking on an option in layer one, the layer two will show these options and hide the options of layer one. This process works for all the scenes. If you clicked on an option and there is no further layer to show, the dialogue is completed. Corresponding to the answers you have clicked a number sequence is written into an answer string. This number sequence is 32 Chapter 4. Realisation of ‚Projekt: Rose‘ composed of the given answers. Each option in the dialogue has a number according to the line in the text-file where it is stored. Black Screen To finish the whole dialogue process a black screen will be displayed at the end. It contains a text that represents the consequences after you clicked the last option. This consequence is independent from the chosen options. The black background is a simple black image with a small size attached to a canvas 8 in the game. In the script FadeScreen it is initialized in the start method 9 and it is set to false at first. That means it is not visible to the test person. Just before the next scene will be loaded after a dialogue was completed a method sets the visibility to true. Life Display The life display was inserted to give feedback to the test persons decisions. Any time a player will finish a scene dialogue, the life display will be updated. But independent of the decisions the consequence is always negative, therefore the displayed bouquet of roses get damaged: F IGURE 4.8: Rose Life Cycle (from left to right) In the game it is realized as a small canvas containing a script. In addition, there are five images which are predefined and located in the assets folder. The script RoseHealth has only a start method that will actualize the image in the canvas each time a new scene will be loaded. In addition, the image only changes if a player has completed the dialogue of the scene. But if you just leave a scene and move to the next one, the rose image doesn’t change. 4.6.6 Data Collection To evaluate the data after all the test persons has completed the simulation a data collection was built-in. It gathers the required information and write them into a CSV file. The process starts with the completion of the first scene dialogue. A number sequence will be written into a answer string. If the next level will be loaded the answer string is written into a string array as well as the number of the scene, the input value of the test person for the number sequence, the date, the start time, the end time 10 , the position of the actual scene in the number sequence. The string array containing the answers is the last value. These data are stored first in a two dimensional array. It 8 Corresponding to the Unity API, a canvas is an “Element that can be used for screen rendering. Elements on a canvas are rendered AFTER scene rendering, either from an attached camera or using overlay mode.” 9 That means if you start a scene and a game object with a script containing a start method is in this scene, the method will be called at the start. 10 All three time data are in the format of the Unix Timestamp, where only the start and end time also capture milliseconds. 4.7. Discussion on Predictability 33 is constructed that the first value represents the position an information is sorted to and the second value represents the information itself. Afterwards the two dimensional array is used to join the information into one string and separate the information with a delimiter. The whole string will be written into an end string that is stored constantly until the game ends. The described method is repeated after each scene. At the end of the whole simulation the string will be written into a CSV file. The first step is to create a desktop folder to store all the CSV files. For all future test persons it is proved if this folder already exists or not, to avoid creating the folder each time or overwriting it. Beyond, the strings of the end array are stored into the CSV file where the strings are ordered by the scene number. A condition proves if the CSV file to write already exists and if it is not existent the CSV file is saved to the folder created before. The name of the CSV file is composed of “Subject_”, the scene sequence and the actual date. To guarantee that the CSV file can be created a sign within the calculated time will be replaced by a “_”. 4.6.7 Scene Skipping If a test person starts a scene he or she always has the option to leave the scene. There is a predefined border in each scene. If the test person move into the border a message will show up. The message contains the question “Möchtest du die Szene wirklich verlassen”. Optional he or she can press “Ja” or “Nein”. If a player clicked on the “Ja” button the answer string will contain “skipped”. A script named BorderNotification is attached to the ingame border. It contains a real simple method that is called as soon as a test person enters the trigger of the border. The method draws a GUI label with the question mentioned before and two buttons for each possible answer. The time scale was set to zero to avoid malfunction of the simulation if a test person would decide to walk back to the scene and just click on a character. 4.7 Discussion on Predictability The functions for prediction were not integrated due to a lack of time. But some possibilities for the implementation were intended. Eye- and mouse tracking were considered as useful techniques to observe a player during game playing. Meanwhile some data should be collected, e.g. eye fixation of objects in the 3D environment and the speed of the mouse movement. The analysis of the data should provide informations about the behaviour of the player during the game and especially inside the dialogue processes. It should be determined if there are conspicuous behaviours of the eye and the mouse controlling hand in combination with other collected data. This wide analysis could make assertions about the predictability of behaviour in moral decisions. 35 Chapter 5 Resume & Potential for Future Research 5.1 Conclusion At the beginning of this thesis I presented an overview of studies which contain morality and moral behaviour. Those studies of Hume and Kant have constitute two traditions and influenced succeeding researchers like Haidt and Kohlberg. The developed theories and models should categorize moral dilemmata and lead to a further understanding of moral behavior. Thereafter I have shown some aspects of morality in video games. With the interaction mode and the player behaviour by age I wanted to present studies that include both video games and moral dilemmata. It should give an impression of the present state of the art, except the age group “children”. The study with the children was used to describe the different but necessary approach to explore morality and moral behaviour of children. After this chapter I have illustrated the foregoing process to discover the characters and dialogue options by Ena Franz and Victoria Götz. And furthermore, I have shown the steps for the conception of the game. The processed elements of the game were introduced in which I have described the element as well as the possibilities and the chosen methods. These elements include the surrounding environment and the interfaces and also the character features I have identified as important. At least I have shown the implementation of the game step-by-step. The technical details were put in front of the chapter. They contain the technical Details of the computer, where the game was developed and all used programs. Afterwards, the steps of implementation were presented, starting with the character creation. The procedure was described in detail to establish an understanding. The same approach was applied to all further topics: Character Animation, Scenes and Interfaces. Within 4.6 I have decided to amplify the methods of the scripts, to display the functions behind the visible. 5.2 Proposal for Improvement The developed simulation has much potential for further improvements to explore the moral behaviour and the prediction of moral behaviour.Therefore I’d like to provide some enhancements and extensions for the game. The first step might be the animation of all characters. It could lead to a more vibrant world and maybe stimulates a player for more involvement. One 36 Chapter 5. Resume & Potential for Future Research should mind the natural movement as I already mentioned in 3.2.2. Another valuable feature could be the integration of audio samples for each characters dialogue the player is confronted with and also for the foregoing dialogues between the NPCs. This implementation should be realized in connection with the realization of facial expressions. Therefore you may avoid to confuse the player and get unwanted player behaviour. To examine the predictability of the player behaviour I propose to integrate an eye tracker and mouse tracking scripts. Both scripts can be added directly to the game assets. For the eye tracking I recommend you to use,if possible, an eye tracker device (e.g. from Tobii), because most of them bring classes you can use and adapt for your purposes. If you want to use a webcam or another camera instead, you may have to program the functionality on your own, if the camera supports it. A last improvement I’d like to propose is the adaptation of the environment. Maybe it is possible to evolve an interface linked with the unity editor, to exchange objects of different assets, so you can build a whole new environment. It could work like a drag and drop function, where one side represents the actual assets and the other side uses the explorer. Though one can search for the objects one wants to integrate and easily exchange it with the actual object. It goes without saying that the objects should have a meaningful designation, so one will now, which objects will interchanged. 5.3 Future Research There is still potential for future research in morality and moral behaviour of humans. I think, it can be expected that the changing society might entail a change of morality and moral behaviour, too. Therefore the changes of the understanding of morality would ensure that older theories and models must be revised, or new ones have to be developed. In addition, the technological progress may provide new devices and methods. These might be useful for the research in morality and moral behaviour. 37 Bibliography Borg, J. Schaich et al. (2006). “Consequences, Action, and Intention as Factors in Moral Judgments: An fMRI Investigation”. In: Journal of Cognitive Neuroscience 18.5, pp. 803–817. Danovitch, J. H. and F. C. Keil (2007). “Choosing Between Hearts and Minds: Children’s Understanding of Moral Advisors”. In: Cognitive Development 22.15. Franz, E. and V. Götz (2014). What is right - what is wrong? An explorative analysis of moral reasoning regarding different media. Gentile, D. A. et al. (2009). “The Effects of Prosocial Video Games on Prosocial Behaviors: International Evidence From Correlational, Longitudinal, and Experimental Studies”. In: PSPB 35, pp. 752–763. URL: http: //psp.sagepub.com/cgi/content/abstract/35/6/752. Herz, D. and A. Blätte (2000). Simulation und Planspiel in den Sozialwissenschaften. Eine Bestandsaufnahme der internationalen Diskussion. Münster, Hamburg, London: Lit Verlag. Jackèl, D., S. Neunreither, and F. Wagner (2006). Methoden der Computeranimation. Berlin, Heidelberg: Springer Verlag. Monin, B., D. A. Pizarro, and J. S. Beer (2007). “Deciding Versus Reacting: Conceptions of Moral Judgment and the Reason-Affect Debate”. In: Review of General Psychology 11.2, pp. 99–111. Prensky, M. (2001). ““Simulations”: Are They Games?” In: Digital GameBased Learning, pp. 1–10. Svelch, J. (2014). “The Good, The Bad, and The Player: The Challenges to Moral Engagement in Single-Player Avatar-Based Video Games”. In: pp. 1–17. 39 Appendix A Downloaded Assets TABLE A.1: Downloaded Assets Folder Name ADG_Textures Apartment_House apartments Bambo_House Bld_11 building1 building2 business1 Business_Building Mall Old_houses ResidentialLite Ampel ChairMetal_01 parkbench playground Computer Pack PBS Materials Flowerbed Free_SpeedTrees hedge maple Flower Set plants Stones_Bush tree Weide QS Roof Textures Sand sky5x Streets Stores texturepackA Fabric Materials Tiles_01 TL_LAB Original Asset Name 10 High Resolution Wall Textures Building N210209 Building N250814 House N090614 Hotel building facade business Building N101213 Building N150509 Building N050510 Building_business_centre N2 Building N240510 AK - Poorfolk’s urban housing Residential Buildings Lite Traffic light N260215 Chair Metal One ParkChair Slide N100315 Computer Font Pack PBS Materials Flowers N240713 Free_SpeedTrees Bush N240713 Maple Tree 4 N090813 Ornamental Flower Set Small Plants Bush stones N230814 Tree Poplar 2 Tree N180315 QS Materials Nature Roof Textures Yoghues Free Sand Materials Sky5x One Simple Modular Street Kit Building N070409 Seamless Texture Pack Yoghues Free Fabric Materials Grey Stone Tiles Grass 01 Game-Ready Source Asset Store archive3d archive3d archive3d artist-3d archive3d archive3d archive3d archive3d archive3d sharecg Asset Store archive3d Asset Store Asset Store archive3d Asset Store Asset Store archive3d Asset Store archive3d archive3d Asset Store Asset Store archive3d archive3d archive3d Asset Store Asset Store Asset Store Asset Store Asset Store archive3d Asset Store Asset Store Asset Store Asset Store 40 Appendix A. Downloaded Assets Further Notes on Asset Folders Editor The “CrossPlatformInput” folder if from Unity and “EyeXClientLibraryDeployer” folder was a component of the Tobii EyeTracker. Leather01 This folder contains materials from the developer “Gullrat” and downloaded from brusheezy.com. NPC The assets and objects in this folder are components of MakeHuman and partially modified by Blender. Standard Assets & Plugins & UnityVS All assets within these folders are components of Unity. Wool This folder contains materials from the developer “lostsync” and downloaded from brusheezy.com. Other Assets All of the other assets which are not mentioned above were created by myself. Comment on some Assets Please note that two assets are no longer available in the Asset Store: Residential Buildings Lite and Chair Metal One. There is the following comment in the Asset Store: “Unfortunately, [name of the asset] is no longer available. This package has been deprecated from the Asset Store. This means that new purchases of the package are not allowed and that only users who already purchased or downloaded the package before it was deprecated, are allowed to download it. In most cases, package deprecation happens because the publisher is unable or unwilling to support the package anymore. We suggest looking for alternative packages or contacting the publisher directly. If you’ve already purchased it and need to download a copy, you can do so here.” 41 Appendix B Content of the DVD The DVD contains a .rar archive with the following content: • a folder that contains all files from the game, indlcuding the .exe file • the “ReadMe” file with some additional information that one might notice, if he or she wants to improve the game • the Master’s thesis in PDF format. 43 Appendix C Scenes The following screenshots represent the dilemmata in the same order as they were implemented in the game. F IGURE C.1: Scene 1 44 Appendix C. Scenes F IGURE C.2: Scene 2 F IGURE C.3: Scene 3 45 Appendix C. Scenes F IGURE C.4: Scene 4 F IGURE C.5: Scene 5 Erklärung Ich erkläre hiermit gemäß §17 Abs. 2 APO, dass ich die vorstehende Masterarbeit selbständig verfasst und keine anderen als die angegebenen Quellen und Hilfsmittel benutzt habe. Bamberg, den 28. April 2016 ______________________________ Unterschrift