To game or not to game: examining differences in presence, arousal and intended aggression after playing or observing a violent videogame
P. van den Broek
Department of Psychology
Faculty of Behavioural Sciences
University of Twente
Enschede, The Netherlands
You are about to read the result of a study I conducted to achieve my Master of Science degree in Psychology at the university of Twente, Enschede, The Netherlands. Although the past months were very busy because finding a job required more of my attention than I had anticipated for, I can look back at a very enjoyable period. With the lingering graduation period for my previous degree still in my mind, it was refreshing and pleasant to find two supervisors who gave me the support I needed, both with respect to the content of my thesis as well as practical affairs surrounding the experiment itself. Therefore, my compliments as well as my gratitude to Oscar Peters, my first supervisor and Ard Heuvelman, my second supervisor.
Second of all I would like to thank Peter Moor, who was so kind to replace me as the experiment leader for the second session of the experiment, without you I would have had a big problem. I will not individually thank all the participants of the course Media Psychology who made this experiment possible, needless to say, without you there would be no experiment and no thesis.
Thank you friends and family for your interest in my work (this time I will not thank you for listening to my worries since I barely had them). Finally, thank you Marleen for your support and critical input in the times progress was slow or when I just wanted confirmation of my ideas.
Pim van den Broek
Amersfoort, May 2008.
This exploratory study examined the difference in intended aggression after playing (active) and observing (passive) a violent video game. The main explaining variables were arousal and presence, independent variables like trait aggression, playtime, game preference, previous experience with the game and gender were analysed as possible interacting variables. It was predicted that participants in the active condition would score higher on the intended aggression scale than participants in the passive condition through more arousal and presence. Results indicate that participants in the active condition were significantly more aroused and perceived greater feelings of presence. However, these variables were not correlated to intended aggression and no differences on intended aggression were found between conditions. Previous experience with the game, gender and a preference for first person shooter (FPS) games proved to interact with condition, however the difference between the active and passive condition was only significant within the group that frequently played FPS games. Moreover, this difference was in the opposite direction as predicted. Although differences within the group that plays FPS games frequently hint toward a difference in intended aggression between the conditions, the results for the explaining variables were too ambiguous and contradictory to draw conclusions from. Until more research is conducted for the specific sub-groups (amongst others FPS players) it is concluded that there is no difference between playing and observing a violent video game.
The influence of (violent) media on its users is controversial (Perse, 2001; Sparks and Sparks, 2002), as well as a hot (research) topic, certainly when another violent drama like the 2002 Erfurt shooting in Germany takes place (where 13 people were killed by a nineteen year old). Some critics and the papers were quick in pointing out a scapegoat: violent media like gangster-rap, television and perhaps most importantly: violent videogame (in this case: Counter strike). Headlines like “software for a massacre” and “the killer was trained by a computer game” leave little doubt about the general attitude that violent computer games cause violent behaviour (Murphy, 2002).
Although there is current research which indicates a fairly strong correlation between violence in real life and playing violent videogames (e.g. Anderson, Gentile and Buckley, 2007; Anderson, 2004), there is also plenty of research that contradicts this (e.g. Sherry, 2001; Ferguson, 2007) or concludes that causation is not easy to indicate at best, mainly because of the many limitations of current research (Freedman, 2001; Cumberbatch, 2004; Boyle and Hibberd, 2005). A discussion of these limitations is out of the scope of this study, for an extensive overview of the limitations of current violent video game studies, also see Rutter and Bryce (2006).
As noted before, a causal relationship between media violence and real life aggression is hard to indicate and has not (yet) been statistically proven. Hence, the use of the word influence of (violent) media above instead of effect, to point out the complex interplay between social and environmental factors and media. This is opposed to the “effects” tradition, where (perhaps erroneously) the term “effects” suggests a causal relation between media and human behaviour which can be studied in isolation (Giles, 2003; Gauntlett, 1995).
This does not hold some scholars back to claim that ‘violent videogames cause real-life violence’, i.e. they treat the causal relation as proven (Bushman and Anderson, 2001; Carnagey and Anderson, 2004). Anderson, Gentile and Buckley (2007, 29) without batting an eyelid even claim that “[...] the evidence (i.e. previous research) strongly points to a significant deleterious effect of violent media on aggression and aggression-related variables”. Other explanations, such as the shooters’ social rejection, feelings of alienation at school, and depression are mostly treated as minor factors compared with video games (Sternheimer, 2007). Ferguson (2007) adds that “[in the discussion of causes of violent drama’s] less than 1% of news stories focused on the responsibility and moral character of the perpetrators themselves”. Gee (2006) further states:
“[...] claiming that video game violence causes real life violence is akin, I suppose, to the claim that because I have planted lots of corn in Harvest Moon I will run out and plant corn in my back yard—in reality we have as little real corn from Harvest Moon as we have real killings from Grand Theft Auto (which is not to rule out the rare case of either—given enough time even low probability events occur—though, of course, by definition, rarely).”
Perksy and Blascovich (2007) studied this idea, only not in the context of planting crops, but in the transfer of artistic and creative feelings when a (non-violent) video game was played which supposedly stimulated these feelings. No results were found, with explanations ranging from “violent content is more readily transferred and intensified” to “the non-violent condition did not engender creative feelings”. Although the results seem to support the words by Gee (2006) and simply banning violent video games from society will unlikely bring down real-life violence, since it is part of a much more complex network of personal and contextual influences, many believe that a ban or censor is the solution.
Banning and censoring violent video games is as a matter of fact at the order of the day. Although Wikipedia is not an official source, an interesting overview of the video game controversy and many examples of censorship and banning in different countries around the world can be found there1.
This extensive overview and the discussed doubtful relation between media violence and real life violence gives rise to the question whether all this attention and preventive action to protect citizens is not exaggerated, especially in the light of more passive forms of violent media which have been around much longer than semi-realistic violent video games, which only developed in the last few years (the pre-2000 games can hardly be compared to for example movies in terms of realistic graphic violence). It is nevertheless a popular belief that media where the user is actively involved in the violence (i.e. video games), have a greater impact on its users than media where users play a more passive role (e.g. television and movies), this active involvement might cause mediated violence to be experienced as if it was non-mediated, it may be more arousing, seem more “realistic,” and be more likely to desensitize viewers (Anderson, Gentile and Buckly, 2007, 75; Bryce and Rutter, 2006, 207; Lombard, Reich, Grabe, Bracken and Ditton, 2000).
This study intends to examine whether this is actually the case, investigating the role of presence (or the conviction of being located in a mediated environment) and arousal on aggression after playing or observing a violent video game. It should be noted however that although this study is connected to most violent video game research, it does not intend to interfere in the discussion whether violent media (passive as well as active) indeed cause violent behaviour. The main objective is to examine possible differences between active participation and passive undergoing of violent video games in terms of induced aggression, presence and arousal without comparing the results to non-aggressive forms of media (which is also not possible given the research design).
The follow-up of this chapter first offers a short overview of the theoretical mechanisms that underlie mediated aggression, followed by a brief review of the General Affective Aggression Model (GAAM) which claims to integrate many existing theories. Finally a short overview of measures of aggression that are used in research today will be given.
Theoretical mechanisms underlying the influence of mediated violence
Media influences can be described along various dimensions (Perse, 2001), some of which will be discussed in this introduction. The first is micro versus macro. The former concerns individual influences whereas the latter is a more societal influence. However, it is a misconception to consider macro level effects are accumulations of micro level effects, as individual influences might differ under the influence of differences in for example education or income.
Another dimension is intentional versus unintentional. To clarify this in the context of commercials the intended influence is obviously to promote something and influence consumers to buy something. However, an unintentional influence when the commercial is repeated too often might be that consumers develop a negative brand image because of irritation. For violent video games the intentional influence is to amuse consumers of the game or serve as an outlet for stress. An unintentional influence is that consumers might start behaving more aggressive in real life as well.
Long term versus short term influence needs little elaboration. The most important aspect here is how long certain influences are expected to last. Arousal for instance last relatively short, whereas the learning of social scripts is more a long term process. Most experimental research focuses on short term influences since these can be measured directly after the stimulus and are less susceptible to variables which are hard to control for.
The last dimension which will be discussed here is the way in which media can influence people individually, the first being cognitive: the way information is acquired – what people learn, how this information is structured and how information needs are satisfied or not. The second dimension is affective – the formation of attitudes, feelings and (positive/negative) evaluations. The third dimension comprises behaviour – observable actions which are potentially elicited by media exposure.
Besides the dimensions of influences, there are several theoretical mechanisms that underlie (i.e. explain) these influences. These theories which are often cited in literature about media effects will be discussed subsequently (Dill and Dill, 1998):
Social Learning / imitation
Excitation transfer / arousal
This mechanism, based on Albert Bandura’s social learning theory (e.g. Bandura, 1971), states that behaviour is learned through observation and imitation of attractive, rewarded models (Sherry, 2001). Much of this learning takes place without an intention to learn and without awareness that learning has occurred (Anderson et al., 2003).
Observational learning can help to explain some of the short-term influences of exposure to violent media when observed behaviour is immediately imitated. Long-term influences on behaviour (i.e. persistency of the behaviour) are first of all dependent on reinforcement from the environment. Reward or punishment influences whether the imitated behaviour will recur in the future. A second way aggressive behaviour can be stimulated is by disinhibition; with age and development, internal impulse controls are created to inhibit aggressive behaviour. These controls can be weakened when aggressive acts are observed (Calvert and Tan, 1994).
Besides the direct imitation of behaviour of models in the media, it is also possible that certain social scripts are learned, which are a guideline for future behaviour (e.g. how to respond to conflict), which can be considered a long-term influence. When violent models are observed in the media, the user of the media might learn (or come to believe) that violence is a good or normal way to solve a conflict.
When the mechanism of social learning is evaluated in the context of violent video games, it predicts that users could translate the aggressive behaviour of their video game character to real life conflict situations because it is rewarding in the game situation (e.g. the player advances in the game or gets bonus points when using violence). From this point of view violent videogames can be considered a simulation of real-life violence and the danger of violent video games that the simulated behaviour will be brought into practise, which makes them potentially more dangerous than television or music which are passive instead of active media (Dill and Dill, 1998). As noted in the introduction, this explanation is often used by the media to point out a scapegoat, in the case of violent excesses in our society.
Arousal and Excitation transfer
Emotions can be described in terms of two basic dimensions, arousal and valence. Arousal is the intensity of the emotion, ranging from feelings of being energized, excited and alert to feeling calm, drowsy and peaceful (Reeves and Nass, 1996, 132). Valence is the direction of the emotion, good or bad, likeable or not likeable, worthy to approach or something to avoid. All arousing experiences have important consequences, in the case of a negatively valenced experience it helps activating primitive fight or flight responses and in the case of positive valence it gives an opportunity for pleasure or procreation.
Arousal can influence aggression in four ways. First, arousal from an irrelevant source can increase the vigilance with which people observe the world and it strengthens current behavioural tendencies. In the case that a person has aggressive feelings or thoughts, high arousal can be the enabling factor to act out these feelings and behave aggressively. If a person is provoked to aggress at the time that increased arousal occurs, heightened aggression can result (Reeves and Nass, 1996; Anderson and Bushman, 2002).
Second, arousal plays an important role in the theory of excitation transfer (Zillman, 1971), which takes place when adrenaline produced by an exciting stimulus carries over to a later activity which may be mislabelled as the cause of the excitement (Reeves and Nass, 1996). In the context of media violence, an exciting mediated experience might result in an excess of arousal. Successively a person gets into an argument with someone else which results in physical confrontation, unaware that the physiological excitation is the residue of the mediated experience and not caused by the argument itself. However, this seems only plausible when the period between the stimulus and the other activity is not too long (i.e. as arousal fades after the stimulus, so does the chance for excitation transfer) (Giles, 2003). Excitation transfer suggests that the arousal effect may also persist over a longer period of time: even after the arousal has dissipated, the individual may remain potentially aggressive for as long as the self-generated label of “angry” persists (Anderson and Bushman, 2002).
However, there are some limitations concerning the source of the arousal; when arousal is elicited by sources antagonistic or unlike an aggressive provocation, it is less susceptible to transfer. The reason for this may be that for instance pleasant experiences also produce feelings and thoughts opposite to anger and the likelihood that a person will misattribute euphoric arousal to provocation is small (Tyson, 1998).
Moreover, Sherry (2001) states an inverse effect of excitation transfer: although a stimulus might be initially exciting, arousal might be replaced by fatigue or boredom after prolonged exposure to the stimulus (e.g. long game play sessions). Excitation transfer would then predict a reduction in aggressive tendencies.
Third, arousal can have another long term influence on aggression: highly arousing experiences (mediated or not) are associated with better memory for the experience (Reeves and Nass, 1996), possibly contributing to the internalisation of aggressive scripts in the context of social learning .
A fourth and as yet untested possibility is that unusually high or low levels of arousal may be aversive states (i.e. an internal state which is unwanted), and may therefore stimulate aggression in the same way as other aversive or painful stimuli (Anderson and Bushmann, 2002). Zillmann (1983) argued that people experience a cognitive incapacitation when highly aroused, reducing inhibition mechanisms (e.g. situational influences) and increasing the probability of well-learned responses (which might contribute to the display of violent behaviour learned through violent media).
A large number of situational variables influence both physiological and psychological arousal. Exercise increases both, whereas alcohol decreases both. Interestingly, changes in physiological and psychological arousal do not always coincide. Note that this is in important implication for experimental research where both measures of self-report (psychological) and for instance heart rate (physiological) are used to measure aggression. Anderson et al. (2000) for instance found that hot temperatures increase heart rate while simultaneously decreasing perceived arousal. This suggests that heat also might increase aggression through physiological arousal, but this results would not have been found for self-reported arousal.
Whether intentional (e.g. therapeutic desensitization) or unintentional (e.g. by observing media violence), desensitization allows people to become less sensitive (or even apathetic) to certain aspects of their environment. This can be a positive development (e.g. surgeons becoming resistant to distressing sights) or negative (people becoming less sensitive to violence in society). The term “desensitization” in the context of violence has many meanings, amongst others a reduction in physiological arousal or affective reactions to real life violence or a reduction in sympathy for a violence victim. Therefore, Carnagey, Anderson and Bushman (2007) suggest a narrower definition of desensitization: a reduction in emotion-related physiological reactivity to real violence.
Anderson et al. (2003) state that the diminution of unpleasant physiological arousal (or negative emotional reactions) as an influence of much exposure to media violence is associated with a heightened likelihood of violent thoughts and behaviours. The underlying theory is that these negative emotions normally have an inhibitory influence on thinking about violence, condoning violence or behaving violently, overexposure to mediated violence diminishes or annuls this effect.
As opposed to the previously discussed mechanisms, catharsis actually predicts a positive influence of violent media on its users. It states that the consumption of violent media offer a way of discharging aggressive feelings by observing (or participating in) a fictional act of violence. In the context of violent video games catharsis provides a way of acting out aggression that is not allowed in the real world (Sherry, 2001). In the psychoanalytic theory this is also called aggressive drive reduction, where violent video games offer a way to “drain off” these impulses (Calvert and Tan, 1994).
Catharsis has many historical interpretations, but, as applied to anger and aggression, it was a hypothesis about conservation of emotional energy (Tyson, 1998). Anger or aggression that is once provoked must be expressed in one form or the other. However, in a study where people were provoked and were given the opportunity to aggress against the experimenter afterwards, a decrement in arousal was only found in real life aggression (e.g. physical or verbal) towards the experimenter, not for fantasy attacks (Tyson, 1998, 148). This implicates that playing an aggressive video game does not lower arousal.
Priming theory posits that cues from the environment may lead to aggression or hostility due to the priming of semantically related informational nodes (Berkowitz and Rogers, 1986). Or a less formal description: in this approach, concepts that are related (e.g., river and water) can trigger one another and perhaps related behaviours such as thirst and drinking (Farrar and Krcmar, 2006). When this is applied to violent media: when a violent scene is observed, this may activate (or increase the accessibility of) a complex set of thoughts, feelings and behavioural scripts which are related to aggression (Anderson et al., 2003). Because these schema’s are more easily available, an interpretational bias is created, which can increase the likelihood that a person will react aggressively in an ambiguous situation, for instance when bumped into on the street.
A long term influence of the priming mechanism strongly relates to the concept of cultivation, where media shape people’s beliefs about the real world. A striking example is the so called “mean world”-effect; people seriously overestimate the amount of crime in the real world because of large-scale media coverage of violence in our society (Giles, 2003, 21-22), which results in a general climate of fear or unsafety (Van Mierlo and Van den Bulck, 2004).
The General Affective Aggression Model (GAAM)
This model by Anderson and Dill (2000) claims to integrate the influences on the cognitive and the affective dimension into a model for predicting aggressive behaviour after exposure to violent media. Many of the earlier discussed theoretical mechanisms can be recognised, inhibition and desensitisation for instance can be characterised as personal variables, arousal is integrated as a present internal state variable, priming plays a role in present cognitions and affects and (the potential ability to induce) presence can be characterised as a media variable (although presence is also dependent on personal variables and someone’s internal state). The model is depicted in Figure 1 on the next page.
The input variables consist of personal and situational variables which influence the internal state of a person. The personal variable for instance encompasses how easily aggression related information is accessible by nature (which can be influenced by previous experiences, so nurture for example by social learning and desensitisation also plays an important role). This variable can for instance be measured by assessing the person on the Buss-Durkee Hostility Inventory (BDHI) (Buss and Durkee, 1957). Situational variables can also influence the accessibility of aggression related knowledge structures, for instance when a person is treated unfair. Anderson and Dill (2000, 772-774) believe that playing a violent video game is such an external factor too (although they do not believe that routinely playing video games increase feelings of hostility or anger).
As said before the input variables influence cognitions and affects (and each other) and arousal, which in turn are influential in the likelihood that a person will react aggressively to a certain situation through the earlier discussed theoretical mechanisms.
In the end, the current internal state influences what behaviour expressed, which might be automatic or controlled (as depicted by the appraisal process). This behaviour in turn might reciprocally influence the situation the person is in (a situational variable): when the person under consideration starts calling names, the argument might escalate, he/she could be hit by the target after which he/she might hit back.
Figure 1: Schematic depiction of a single episode of the GAAM model
Measurement of aggression
The measurement of aggression is as controversial (i.e. the validity of current aggression scales) as the question whether violent media (or more specifically violent video games) cause aggressive thoughts, feelings and/or behaviour (e.g. Freedman, 2001). The cognitive influence of violent video games on participants is typically measured using reaction time tasks, for instance how quickly (aggressive) words are read aloud from a display (Anderson and Dill, 2000) or the amount of interference caused by certain (aggressive and non-aggressive) words on a modified Stroop task (Kirsh, Olczak and Mounts, 2005). Another method is the way in which participants finish ambiguous incomplete words after exposure to different types of media (Carnagey and Anderson, 2005), for instance H_T which can be completed as HIT (aggressive response) or HAT (non-aggressive response). These scales rely the priming theory to explain the cognitive influence of violent media.
The influence on the affective dimension is mostly measured using self-report scales, for example the State Hostility Scale (Anderson and Dill, 2000, 784) where participants are asked to rate statements like “I feel angry” and “I feel mean” on a 5 point Likert scale where 1 means “strongly disagree” and 5 means “strongly agree”. The underlying theory is that violent videogames (but also violent media in general) have a priming effect, which activates hostile thoughts, associations and feelings (Giles, 2003, 66).
Another method is physiological, for instance by measuring Facial ElectromyoGraphic (EMG) activity, for instance over muscles that control the eyebrows and the angle of the mouth (Ravaja, Turpeinen, Saari, Puttonen, Keltikangas-Järvinen, 2008). These EMG’s can then be used to index the valence of emotions. Furthermore, Electroencephalography (EEG) can be used to study cortical responses to different stimulus events (e.g. violent and non-violent), where different EEG frequency bands have been associated with various processes, amongst others emotional responses (Salminen, Ravaja, 2008).
Recently researched behavioural influences include pro-social behaviour as represented by decision making in the Prisoner’s Dilemma game (Sheese and Graziano, 2005) and the Taylor Competitive Reaction Time (TCRT) task. In the latter method, the participant's goal is to push a button faster than his or her opponent. When participants lose this race, they receive a noise blast at a level supposedly set by the opponent (actually set by the computer). Aggressive behaviour is defined as the intensity and duration of noise blasts the participant chooses to deliver to the opponent (Anderson and Dill, 2000; Carnagey and Anderson, 2005; Bushman, Ridge, Das, Key and Busath, 2007).
Although behavioural measures seem compelling, Farrar, Krcmar and Nowak (2006) state that they are problematic in at least several ways. First, they are more difficult and time consuming to administer than the typical self-report measure. Second, special equipment is typically needed that can be both costly to obtain and time consuming to use, for instance in the case of the TCRT task as described above. Third, despite their seeming external validity, one obvious problem is that the more valid the behavioural measure appears, the less ethical it is to use. Finally these methods mostly take place in a laboratory or an artificial setting which might reduce the external validity of the results, which might not be representative of “real world” violence (Giles 2003, 30; Gauntlett, 1995).
Therefore, a sub-aspect of aggressive behaviour is posed: hypothetical or intended aggressive behaviour, which can be measured by confronting participants with hypothetical situations which might result in conflict and asking participants how they or the person in the situation will react (instead of actually asking participants to react). A method to measure intended aggression by Bushman and Anderson (2002) is the ambiguous story stem protocol in which participants have to indicate what the main character will do or say, think, and feel as the story continues (each story ends with “what happens next?”). A less open-ended (i.e. closed instead of open answers) and more personal (i.e. how would you react instead of how would the main character react) version of this protocol was used by Wei (2007) where participants had to indicate how they would react in possible conflict situations choosing from three multiple choice options ranging from physical force to yelling and finding a peaceful solution or doing nothing.
It is a popular belief that media where the user is actively involved in the violence (e.g. video games), have a greater impact on its users than media where users play a more passive role (e.g. television and movies), this active involvement might cause mediated violence to be experienced as if it was non-mediated, it may be more arousing, seem more “realistic,” and be more likely to desensitize viewers (Anderson, Gentile and Buckly, 2007, 75; Bryce and Rutter, 2006, 207; Lombard, Reich, Grabe, Bracken and Ditton 2000).
However, there is little research that directly compares these two, examples are those by Calvert and Tan (1994) who compared playing in a virtual reality environment with observing the same game play on an external monitor or Tamborini et al. (2000) who compared virtual reality with a traditional desktop environment. However, in both studies conditions were quite different, as this study is interested in comparing a desktop platform with observing the same stimulus (and extrapolate the results to for instance watching television).
One of the limitations of current video game research is that very often incongruent stimuli are used (e.g. two totally different game genres or a slow paced game versus a fast paced game, possibly causing third variable effects, Klimmt, 2001). An example is a study by Anderson and Dill (2000) who used the games Myst (non-violent) and Wolfenstein 3D (violent), which are two totally different games (calm puzzle and fast paced first person shooter). It is presumed that the same limitation holds when playing violent video games is compared to for instance watching a violent TV fragment or playing on a different game platform. Of course it is possible to try and control for the differences by means of a pre-test, but for this study this is no solution since we want to know whether there are for instance differences in induced arousal. Therefore, this study is designed around the same stimulus, namely a violent video game, comparing participants at the controls (the actual video game players) with participants who are just spectators (the passive group). Then, both groups can be compared in terms of aggressive intent. This results in the following main research question:
RQ: Is there a difference in aggressive intent after playing and observing a violent video game? As noted before, the main difference between videogames and other media is its interactivity which might result in feelings of non-mediation. This concept of non-mediation is strongly related to the concept of presence, which is not so much a mechanism or theory about the influence of media, but it might play an important (mediating) role. Witmer and Singer (1998, 225) define the concept of presence as “the subjective experience of being in one place or environment, even when one is physically situated in another”, while Lombard and Ditton (1997, 12) define presence as “the perceptual illusion of non-mediation”.
According to Witmer, Jerome and Singer (2005), presence in a mediated environment is influenced by several factors, including: the fidelity of its sensory components, the nature of the required interactions and tasks, the focus of the user’s attention/concentration, the ease with which the user adapts to the demands of the environment and on the user’s previous experiences and current state.
Overall, there are two concepts that explain presence (Witmer, Jerome and Singer, 2005; Eastin and Griffiths, 2006).
The first is involvement, defined as “a psychological state experienced as a consequence of focusing one’s mental energy and attention on a coherent set of stimuli or meaningfully related activities or events”. Involvement with media is increased by offering the user an experience that stimulates, challenges, and engages the user either cognitively, physically, or emotionally.
The second aspect is immersion; “a psychological state characterized by perceiving oneself to be enveloped by, included in, and interacting with an environment that provides a continuous stream of stimuli and experiences”. Immersion in a virtual environment (VE) is reduced by distractions from the environment (e.g. light flashes or sounds) and is increased by factors that facilitate direct interaction with the VE and the performance of VE task activities. An example is perceiving oneself as moving inside a simulated environment or directly interacting with other entities in that environment (Witmer and Singer, 1998).
Lee (2004) discerns between three different types of presence humans can experience. Social presence refers to the experience of virtual actors as if they are real actors or the experience of virtual objects that manifest humaneness. Next, self-presence is described as the experience of a virtual self as the actual self (which might lead to an awareness of themselves inside a virtual environment).
The most interesting form of presence in the context of this study is spatial presence, whichis described as the sense of being physically located in a virtual environment or experiencing virtual objects as if they are real objects or as Wirth et al. (2007, 495) state: “The main characteristic of Spatial Presence is the conviction of being located in a mediated environment”. Note that this definition closely resembles Witmer and Singer’s (1998) general definition of presence. However, Wirth et al. (2007) note that this definition emphasises the immersive properties of the medium too much and passes over the internal processes of the user. They state that even if immersive impulses are not provided by the medium, internal processes like imagination can compensate for the deficit. Therefore, spatial presence is not limited to media using sensory rich technology (e.g. VR), but can also occur when using less immersive media (including text based media).
Wirth et al. (2007) propose a two dimensional construct of spatial presence, with the sensation of being physically situated in the virtual environment at the core (coined as “self-location”). The second dimension refers to the possibility to act based on mental representations of the mediated space: an individual who is experiencing spatial presence will perceive only the action possibilities of the mediated space and will not be aware of (the possible actions of) the real environment. This dimension is called “possible actions”. In this thesis, the term presence will refer to the definition of spatial presence as defined by Wirth et al. (2007).
Involvement is often considered to be a part of the experience of presence. Wirth et al. (2007) see presence as a side effect of media involvement. They conceptualise involvement as “the active and the intense processing of the mediated world”, whereas the concept of (spatial) presence refers to the experience of “being in the mediated world”. However, focussing one’s attention to the mediated environment does not necessarily mean a loss of contact with the real world. Furthermore, if users are for instance involved in the narration of a film or role-playing computer game, they think about the characters, their actions and feelings, but they do not pay attention to the television receiver or the computer mouse in their hands. Wirth et al. (2007) therefore theorise that presence is more a side-effect of higher involvement because the attribution of mental capacity to the mediated world comes at the expense of mental capacity for the real world.
According to these authors, spatial presence is established in stages. First of all it is necessary that a user has a mental model of the mediated environment called a Spatial Situational Model (SSM) and that the user focuses his/her attention to the medium (Attention Allocation). Then, a so called Egocentric Reference Frame2 (ERF) is constructed from visual information, but also other sensory input like vestibular cues. All perceived objects, including one’s own body, are located in reference to the ERF. Thus, an ERF tells us where we are in a spatial environment. Mediated environments may also offer ERFs, for example, in first-person shooter video games (Schneider, Lang, Shin and Bradley, 2004). However, sensory cues can indicate different ERFs when for instance movement through a 3D environment is visually perceived but vestibular cues (which are not addressed by the game) do not indicate movement. To avoid confusion, users are likely to try to return to the condition of one congruent ERF. This is called the Primary Ego Reference Frame (PERF); the user will prefer to align perceptions and action possibilities with this reference frame. Wirth et al. (2007, 506) state that “Spatial Presence occurs when a person accepts a mediated environment as PERF, because in this case, perceived self-location, perceived possible actions and mental capacities are all bound to the mediated space.” Returning to the role of presence in video games, it is mostly expressed in terms of interactivity and interactivity/ active involvement as noted in the first paragraph of this chapter. Interactivity expresses itself in the player’s possibility to select a game character which he/she likes the most (in the case of third person video games), or to experience the game as themselves (i.e. players experience the video game world through the window of their own eyes). Most violent games that are played from this perspective are called First Person Shooters (FPS). Furthermore, the possibility to manipulate or influence the video game world strengthens the perception of a real world (i.e. the perception of “being there”). To investigate whether people actually do feel more like “they are there” when playing a video game, the following hypothesis is proposed:
H1: playing a violent video game induces greater feelings of presence than observing a violent video game. As noted before, the impact of violent videogames is presumed to be greater than that of more traditional (violent) media, mainly because of their interactive character. Explanations include more identification with the game character and the mere fact that the user is actively involved instead of an observer, contribute to greater levels of presence.
A study by Farrar, Krcmar and Nowak (2006) found that presence in violent videogames is positively correlated to hostile intentions. Other studies like those of Calvert and Tan (1994), demonstrated that playing a violent game using a Virtual Environment (VE, also known as Virtual Reality, VR) led to more aggressive feelings than observing game play. Although the concept of presence is not explicitly considered in their study, Persky and Blascovich (2007) found that the use of VE (which is presumed to induce more presence than traditional, non-VE environments) led to more self-reported aggressive feelings and more behavioural aggression than playing on a traditional desktop computer platform.
However, a study by Tamborini et al. (2000) found an unexpected trend toward increased presence and hostility in a desktop computer condition over a VE condition. Moreover, Tamborini et al. (2004) found some support for the connection between presence and aggression; however, overall levels of presence did not predict aggression.
Although current research results about the relation between presence and aggression are inconclusive or even contradictory, the following hypotheses are proposed to investigate the influence of presence on aggressive tendencies:
H2a: greater feelings of presence during violent video game play lead to more intended aggression.
H2b: greater feelings of presence during the observation of a violent video game lead to more intended aggression. Since previous research on the relation between presence and aggression is not conclusive and since there appears to be no previous research on this relation for the observation of a violent video game, the hypothesis is split in two to individually examine the relationship between these two variables for each condition.
As was discussed in the previous chapter, another construct that plays a key role in the influence of violent media is arousal. Calvert and Tan (1994) found a greater physiological arousal for participants who actively played a video game using VE and participants observing game play. To investigate whether the same holds when traditional game play is compared with observation on the same platform, the following hypothesis is proposed:
H3: playing a violent video game induces more arousal than observing a violent video game. Arousal is very often related to aggressive and violent behaviour through mechanisms of excitation transfer, disinhibition and strengthening of behavioural tendencies (also see the previous chapter). Anderson, Deuser and Deneve (1995) found an indication that arousal induced by high temperatures may increase aggression, confirming the potential mediating role of arousal on behaviour (as conceptualised in the GAAM). Furthermore, physiological arousal has been shown to increase learning from media (Schneider, Lang, Shin and Bradley, 2004), which in turn should make aggressive concepts more accessible, facilitating aggressive behaviour. Barlett, Harris and Bruely (2007) found that the amount of blood in the game Mortal Kombat was positively correlated to the amount of physiological arousal and state hostility. Finally, Boyle and Hibberd (2005, 4) find that “[...] There is a body of evidence that playing violent video games increases arousal and the possibility of aggression in some players”. Therefore:
H4a: more arousal leads to more intended aggression after playing violent video games
H4b: more arousal leads to more intended aggression after observing violent video games Although previous research on the relation between arousal and aggression seems more conclusive, the hypothesis is split again, most of all to be able to study the relation in the context of the observing group.
Another interesting relation is the one between presence and arousal; media researchers have predicted that immersive media (i.e. which induce a lot of presence) increase arousal levels and aggression because users feel they are directly experiencing the violent events. Besides, presence seems to be closely linked to attention (or involvement), which in turn is connected to arousal (Dillon, Keogh, Freeman and Davidoff, 2000). Furthermore, it is theorized that highly immersive experiences may influence involuntary (aggressive) emotional responses (Schneider, Lang, Shin and Bradley, 2004), so the relationship between presence and arousal will also be scrutinized.
Furthermore, there are more variables at work. The earlier discussed GAAM model (e.g. Anderson and Dill, 2000), also indicates personal factors, where an aggressive personality seems the most prevailing. It is trivial that a more aggressive person is more likely to have aggressive thoughts and feelings than a person with less disposition to aggression (Bryce and Rutter, 2006).
In their evaluation of the Spanish version of the Aggression Questionnaire (AQ), a scale to measure trait aggression, which will be discussed later on), Santisteban, Alvarado and Recio (2006) found a positive correlation between the score on the AQ and watching television and playing videogames, while extra-curricular reading or moderate amounts of homework were negatively correlated. Measurement of trait aggression allows to analyse the impact of the different conditions on participants with a different disposition to violence (Ravaja, Turpeinen, Saari, Puttonen, Keltikangas-Järvinen, 2008).
Other personal variables that might introduce a third variable effect are frustration and pleasantness of the experience. When for instance a participant does not have any experience with video game play and has troubles with the controls or dislikes the game, he or she can become frustrated or irritated. This can introduce third variable effects on the evaluation of aggressive tendencies and the comparison of playing and watching video games, for instance trough more arousal. However, pleasantness of the experience can also have a positive influence (i.e. less aggressive tendencies) when participants indicate they had a good time (i.e. a catharsis-like effect). Other factors include gender and previous experience with videogames. No hypotheses about these variables are proposed but their interaction with the conditions will be investigated.
Returning to the research question, a deductive approach has been used following the train of thought that playing a violent video game induces more arousal and presence. Since presence and arousal are connected to more aggression in general, it could be that playing a violent video game leads to more intended aggression than observing the same game. However, no hypothesis is proposed about this since the difference in arousal and presence for both conditions (H1 and H3) has not yet been examined. Furthermore, the relationship between presence and aggression in previous research is not conclusive, making inferences about differences in aggression after playing or observing beforehand not sensible. This study intends to explore the differences in aggression between playing and observing a violent video game, focusing on presence and arousal as the most important explaining variables. However, other variables like trait aggression and gender are also taken into account as interacting variables.
To examine the proposed hypotheses and research question, an experiment consisting of two sessions was carried out. Beforehand, a pre-test was conducted to assure the reliability of the scales and to prevent trivial problems with the actual experiment. The participants, stimuli and apparatus and the procedure (for the pre-test as well as both sessions of the actual experiment) will be discussed subsequently.