Being a victim of videogame aggression and your social cooperation (Rothmund et al., 2011)

Gary Smith (Bully)

I’ve reached my milestone of getting through my first year of graduate school and I’m looking forward to something even worse next year: a thesis dissertation. Here are the things I learned during that year: manage your time more effectively, fail the undergrads and save humanity, when I’m in grad school I lost a lot of videogame capital. On the other hand, I’ve gained academic and otaku capital. You make high brow statistics jokes and finally, never enter graduate school unless you are crazy.

Tobias Rothmund (University of Koblenz-Landau) and two colleagues have published an interesting article in Personality and Social Psychology Bulletin. Two experiments that examined how virtual aggression perpetrated by an NPC to the player’s avatar and how it affected participants’ social cooperation with a stranger.


Two experimental studies were used to investigate how interacting with aggressive virtual characters in video games affects trust and cooperation of players. Study 1 demonstrates that experiencing virtual aggression from a victim’s perspective can impair players’ investments in a subsequent common goods dilemma situation. This effect is mediated by reduced expectations of trust in the cooperativeness of interaction partners. In Study 2 the same effect was replicated by using a different cooperation task and by investigating the moderating role of justice sensitivity from a victim’s perspective as a dispositional factor. Participants transferred less money to an unknown partner in a trust game after exposure to aggressive nonplayer characters in a video game. This effect was stronger for people high in victim sensitivity. Results of both studies can be interpreted in line with the sensitivity to mean intentions model and add to the body of research on violent media effects.

I didn’t realized how long it took to recover from the end-of-quarter craziness, it’s been about a month since the last post.

Rothmund and colleagues addressed some issues regarding how videogame violence affects our aggressive and prosocial behaviours. From previous studies, having hostile expectations and attributions is a mediating factor between violent media and aggression. So, if you think a guy who seemed to have pushed you because he’s an asshole or he doesn’t like you, well you are more likely to be aggressive, particularly after a bout of violent videogame play. However, do these same hostile thoughts affect prosocial behaviours as well?

As videogames are interactive media, the player can be both perpetrator and victim of in-game physical violence. Whichever perspective one is taking can influence one’s cognitive, emotional and behavioural reactions. According to the sensitivity to mean intentions model, being victimized can lead a person to become more egotistical and uncooperative to help others because if you have been hurt, you would want to avoid being hurt again in the near future. Individuals who are sensitive of being victimized will tend to be more suspicious when they pick up cues of untrustworthiness and such suspicious mindset leads them towards a more hostile expectations and attributions to others.

The Boss (Metal Gear Solid 3)

Rothmund and colleagues proceeded to hypothesized that the aggressiveness of NPCs can serve as a cue of untrustworthiness and prime suspiciousness, especially if you are directly felt victimized rather than observing being a victim. So, they conducted two experiments. The first experiment examined how perspective of being an active agent or an observer can affect social cooperation. In the second experiment, they examined how victim sensitivity is a moderator in the relationship between NPC aggression and cooperative behaviour.

Study 1


Participants: 100 German male undergraduates, average age is 23 (SD = 2.7). Eight were dropped from the analysis because they were suspicious about the purpose of the study, they were told that study was about something else.


Cooperation: A common goods dilemma game. The participant is given 1 euro and is told they can invest a fraction of it in an investment pool which would be tripled. In addition, they can randomly draw from that pool an investment made by previous participants. The amount of money they kept from their own investment and the random amount from someone else’s investment is what they would receive for their participation. Participants were also asked their opinions about how much the other participants should invest in the pool, this would assess participants’ trust of the others’ cooperativeness. The order of these two questions is randomized.

Risk tolerance: Participants are given 1 euro and are given the option to buy a lottery ticket, if they win the amount is tripled, if not they get nothing. The chances are based on picking a lot of 100 that contained either one of two outcomes, win or lose. Participants were asked the minimum chances of winning that they are willing to buy a lottery ticket.

Video game used: Two scenarios in Bully (Canis Canem Edit in Germany) were used. The low-NPC-aggression condition looks like a training mission where the player is tasked to practice fighting in the Gym. So, in the words of the authors, the player is performing a substantial amount of physical aggression but little aggression by the NPCs. In the high-NPC-aggression condition is the “Help Gary” mission.

I have some reservations about the generalizations from this scenario. First, Gary betrayed Jimmy, translated as relational aggression by the authors and I agree that this is a strong cue of untrustworthiness. Generally, betrayal does not happen often in FPS or RTS games and you don’t see your A.I. friends turn on you in the middle of a level. Second, the betrayal is usually delivered as a narrative device and I believe it is through the narrative rather than the game structure of your typical enemies’ “harm on sight” behavioural pattern is what driving participants’ victimization mindset. Some people could interpret the enemy as loyal people following orders to shoot the player on sight, unless they use “sneak”, “flanking” or “cloak” strategies then I’m inclined to be wary. Third, I cannot equate betrayal with tit-for-tat firefights, even if the latter is when the player is outgunned, it could having different interpretations, such as having a suspicious mindset, challenging mindset or epic fight mindset. Fourth, their measure of aggressiveness is based on the participants’ opinions and the researchers’ content analysis of the acts. Again, the differences lie in the initiation of a combat situation, Gary ambushed Jimmy and by extension the player is different when the player is expected and voluntarily enters combat. They conducted content analyses by counting the number of perpetrator acts versus victim acts, so they are sure that the high-NPC aggression condition has a lot more of Russell pummelling Jimmy than the low-NPC aggression condition. The number of perpetrator acts did not differ between conditions.

According to TV Tropes, this is a Big Bad Friend trope.


Vossler (Final Fantasy XII)

Participants are randomly assigned to either the low- or high-NPC aggression condition and they are either assigned to play or watch the game. The players are given a three-minute tutorial and their playthrough are recorded for the participants in the “watch” condition. After the playthrough or watching, they go through the cooperation and risk tolerance tasks. Participants were told that they would using real money in one of the tasks and that the other participants had played or watched videogames other than Bully because the authors argued that having the same videogame experience could affect trust expectations.


A 2 (game scenario) X 2 (watch or play) X 2 (order of question) ANOVA revealed a single significant two-way interaction. Participants in the “play” “High-NPC aggression” condition invested less money (M = .45 euro SD = .28) and expected that the other participants invested less (M = .42 euro, SD = .18) than those in the “play” “Low-NPC aggression” condition (M = .66 euro, SD = .31) and who expected that the other participant invested a somewhat higher amount (M = .55 euro, SD = .21).

Their analysis in risk tolerance revealed no significant differences between the groups.

Rothmund et al. then conducted a meditational analysis to see the constructs’ (trust and game scenarios) strengths over cooperation. Their analysis revealed that within the player condition, NPC aggression had a smaller direct impact on cooperation and the indirect path is mediated by trust expectation which is stronger. Being betrayed by a friendly NPC impact your cooperation and trust expectations; cooperation depends more strongly to trust expectations. So, if you are still trusting despite the betrayal then it would affect positively affect your cooperation.

Study 2

Experiment 2 used the same procedure with different measures of cooperation and it examined whether victim sensitivity is a personality trait moderating the relationship between game scenarios and cooperation.


Participants: 55 German male undergraduates, average age is 22.4 (SD = 2.6).


The same as experiment 1 except the common goods dilemma tasks are replaced by trust games.

Victim sensitivity: a 10-item 5-point scale. Example: “It makes me angry when I am treated worse than others”. Participants answered this scale 10 days prior to the experiment, so as to arouse suspicion I gather.

Bob (Sin City)

Trust game: Participants are given 1 euro and they were tasked in pooling their one euro in an investment pool along with an ostensible other participant. They were asked how much money they would entrust to another person, hypothetically. Then, the participants were randomly assigned as playing the role of the “truster” or the “trustee”. In the “truster” role, the participants decide how much they would entrust to the “trustee” where the invested amount is tripled. In the trustee role, the amount given by their ostensible “truster” is randomized based on the amount given by previous “trusters”. The “trustee” can keep the transferred money for himself or divide equally between each other. The authors argued the trust game would allow them to disentangle different motivations, which is greed and fear of being exploited.


Their analyses revealed that those in the “truster” role, participants in the “play” “high-NPC aggression” condition (M = .71 euro, SD = .28) transferred less money than those in the “low-NPC aggression” condition (M = .86 euro, SD = .20). However, there were no differences in terms of the “trustee” role, 91.7% and 84% of the respective groups decided to split the investment. The percentage may look like a difference, but the statistical analysis (Chi-square) says otherwise. Their analysis whether victim sensitivity is a moderator revealed that those who are highly sensitive and were in the high-NPC aggression condition have invested much less money than the other groups.


These experiments have shown that suffering an in-game betrayal or high-NPC aggression as the authors argued had a significant impact on how much you trust strangers and your fear of being exploited or generally harmed. Despite the counter-argument that it is only a game and no harm was done, it seems that people do feel something out of it and this is again continued evidence that videogames truly have an impact on us, much like other entertainment media. The authors seem to argue a nuanced approach to cooperative behaviours since those in the trustee role were as likely to share the money with others as the other condition, suggesting that this is not due to general moral desensitization. But, the authors suggested that it might not be since the trustees were given two choices, share or not. A more sensitive measure might accurately rule out the motivation of greed.

The authors argued that taking the victim’s perspective adversely affected trust and cooperation, especially if the person took on an interactive and active role. Again, perspective-taking… it seems logical from a videogame research perspective. However, I would approach it differently and bring “identification with character” into the equation. People who identify with fictional characters, be it videogame or anime are likely to empathize with them through perspective taking and I am wondering what happens if these characters are victimized as well. Do these individuals feel victimized as well as the characters? They should be according to the sensitivity to mean intentions model (I think). Perhaps further studies with character identification might reveal similar effects, maybe Big Boss. Nevertheless, the authors argued that the study increased the understanding of how interactivity can enhance violent videogame effects. IMO, I think active participation just speeds up the process than the slower observation and besides I don’t like Jimmy or Gary. What is also important is that they have identified victim sensitivity as a personality trait and moderator in the relationship.

As I mentioned my interpretation of the study that betrayal, not the aggressiveness of the NPCs is what cued participants’ victimization and suspicious mindset. I am more inclined to interpret aggressive NPCs in the vein of the dehumanization (See Greitemeyer & McLatchie, 2011). People would not view the virtual enemies as social agents, but mechanical agents following attack scripts. The level of trust for dehumanized social agents would be different, perhaps it would not prime a suspicious mindset if they knew they are playing the trust or coop game with a human, but more suspicious if paired with the computer or virtual enemy. What might help is to see if the NPCs were clearly playing unfairly, then perhaps a suspicious mindset would rise and affect social coop or trust tasks. Better dig up some old games and set them to hard…

Some limitations the authors put out: lack of generalizations due to the sample being entirely male, they don’t know how long the effects last, and it doesn’t explain long-term effects. And finally, they have addressed somewhat my criticisms that players do voluntarily play and confront victimization, and how they derive fun from such victimization from aggressive NPCs… Oh wait, I don’t think it does address the betrayal part. And all the characters in this post have one thing in common, they betrayed their friend.

Here’s another thought about the personality trait of victim sensitivity. I read an article in the Columbus Dispatch about trust towards institutions. It made me think if there are any cultural differences of individual and institutional trust. Perhaps Americans are more distrustful than Canadians in sharing their money or signing a contract with Kyubey? Perhaps growing up with such mindset sets a path of immoral motivations, actions and perhaps a more aggressive  (or rather individualistically competitive) stance towards others. Or watching in constant anxiety throughout the series of whether Madoka will sign a Faustian contract with Kyubey.

Related readings: Victim Entitlement to Behave Selfishly (Zitek et al., 2010)

Rothmund, T., Gollwitzer, M., & Klimmt, C. (2011). Of virtual victims and victimized virtues: Differential effects of experienced aggression in video games on social cooperation. Personality and Social Psychology Bulletin , 37 (1), 107-119. DOI: 10.1117/0146167210391103  

1 thought on “Being a victim of videogame aggression and your social cooperation (Rothmund et al., 2011)

  1. Pingback: The game world is violent, I must be careful who to trust (Rothmund et al., 2014) | VG Researcher

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s