A recent study published in Communication Research suggests that online video game players decide a behavior is trolling based on the context surrounding the action, which in turn fuels their desire to strike back. The findings indicate that minor details in a game environment, such as earning points or the general toxicity of the community, shape how players interpret ambiguous actions and whether they retaliate with their own harassment.
Trolling is a persistent issue in online video game communities. Trolls intentionally aim to frustrate, anger, and harm the well-being of other players. However, identifying trolls can be remarkably difficult because they frequently operate within the rules of the game. They exploit loopholes and repurpose normal game mechanics for malicious intent.
Past research indicates that only a small fraction of gamers consistently initiate harassment. Yet, toxic behavior remains widespread. This discrepancy suggests that much of the trolling seen online is actually a chain reaction.
Being a victim of harassment is the most commonly cited reason for a player to engage in the behavior themselves. When gamers retaliate, they often direct their aggression at the original instigator and carry it into future matches with new people. Over time, this contagious hostility becomes woven into the community’s culture.
To understand how this chain reaction starts, scientists look at how people explain the motives of others. According to attribution theories, humans constantly try to reconstruct the reasoning behind someone else’s actions. If a player perceives an in-game behavior as purposefully harmful, they feel a stronger need to retaliate. This retaliation often creates a spiral of reciprocal harassment.
“The study’s motivation is the unchecked spread of trolling in online communities,” said John A. Velez, an associate professor in the Communication Science Unit at The Media School at Indiana University Bloomington. “The main point was to show that players can easily see trolling, even when it’s not there. And that simply thinking they saw trolling is enough for a nice player to spread trolling to others.”
When motivations are not explicitly stated, people rely on environmental clues to guess what another person is thinking. The researchers wanted to identify the specific clues that make gamers interpret an action as trolling.
“I would highlight that it doesn’t matter what a player actually intended when choosing a game behavior; what matters is the interpretation of that behavior by others,” Velez said. “As long as someone thinks they saw trolling, then there is a chance that perception will trigger more trolling in others and spread it over a community.”
To test this, the authors focused on three potential variables. The first was whether the video game rewarded the behavior with points, acting as an external excuse. The second was whether the action targeted a teammate or an opponent. The third was whether the community had a reputation for toxic behavior.
The authors recruited 290 adults from the United States who regularly play video games. The participants were presented with a cover story. They were told they were evaluating real player conduct to help administrators maintain a healthy gaming community.
Participants first read a fictitious code of conduct. They then watched prerecorded, staged gameplay from the popular team-based shooter game *Overwatch*.
Creating a controlled, realistic game environment for an experiment requires extensive preparation. “It took years to complete this study, and my co-authors and I worked really hard to make the gameplay footage; they deserve a lot of credit,” Velez noted.
In the three-minute video, a player controlling a character named Reinhardt engaged in a digital dance, known as an emote, over the body of a recently defeated character. Voice chat and text communication were disabled, meaning the emote was the only way the characters interacted.
The researchers accounted for several variables to ensure accurate results. They measured participants’ prior experience with *Overwatch*. They also recorded the participants’ self-reported history of engaging in toxic behavior and how often they felt targeted by harassment during their own gaming sessions. Controlling for these factors did not alter the pattern of the findings.
The study randomly assigned participants to experience different versions of the video to test the three environmental clues. For the behavioral reward clue, some participants saw text pop up on the screen indicating that the player earned ten points for performing the dance. Other participants watched the exact same dance but saw no points awarded.
For the second clue, the researchers manipulated the group membership of the targeted character. Half of the participants watched the player dance over a defeated opponent. The other half watched the player dance over a fallen teammate.
For the third clue, participants read a fabricated internal report about the gaming community’s norms before watching the video. Half were told that trolling was incredibly common and expected in this group. The other half read that players rarely expected to see toxic behaviors in the community.
After watching the footage, participants rated how they perceived the player’s intentions using a seven-point scale. The data provides evidence that all three clues independently influenced how people interpreted the dance.
When the game awarded no points for the dance, participants rated the behavior as more malicious. On the seven-point scale, perceived trolling scored an average of 4.69 without points, compared to 4.03 when points were awarded. The lack of a point reward seemed to eliminate a game-based excuse for the action, leading observers to assume a hostile intent.
The target of the action also altered interpretations. Dancing over a teammate led to higher perceptions of trolling, averaging 4.60, compared to a score of 4.12 for dancing over an opponent. Players generally expect positive interactions with teammates, so violating that expectation tends to signal malicious intent.
The community’s reputation played a similar role in shaping perceptions. Being told the community was generally toxic increased the perception of trolling to an average of 4.63. In comparison, being told the community was polite resulted in a lower average score of 4.08.
Interestingly, these three environmental clues did not change how the others functioned, but rather piled on top of one another. In statistical terms, the clues were additive rather than interactive.
“The main thing that surprised me was that the effects were additive and not interactive,” Velez said. “Meaning that each sign/auxiliary cue implied trolling separately but that the effect got stronger the more signs/auxiliary cues were included in the observation.”
Researchers initially suspected that one clue might alter the meaning of another. “I originally thought that the signs might change meaning depending on the presence/absence of other signs, but that wasn’t the case,” Velez explained. “It just takes one sign for someone to infer trolling but I would argue there are likely multiple signs in every social interaction and the addition on each sign makes it more likely to see trolling.”
The researchers then analyzed how these interpretations affected the participants’ desire to engage in vigilantism. Participants answered questions about how likely they would be to harass the offending player if they were actually in the game. The options included insulting their intelligence, using slurs, wishing them harm, or intentionally frustrating their gameplay.
The statistical models suggest an indirect chain of events. When environmental clues made the dance look like a malicious act, participants reported a stronger urge to retaliate with their own toxic behaviors. The perception of trolling served as a psychological bridge between the ambiguous action and the desire to strike back.
“The study shows how quickly people make judgments about trolling in a few observations of a player,” Velez noted. “But these types of situations occur frequently in online gameplay and more importantly, each instance can spread trolling to a few more people and then even more people and so on.”
This cycle of misinterpretation and retaliation helps explain why gaming environments can quickly devolve into hostility. “It only takes one misinterpretation to start a cascade of trolling behaviors,” Velez added. “And in the end, the results suggest that the majority of trolling we see online could be troll vigilantes mistakenly trolling other vigilantes. While we can’t stop some people from trolling, we can clear up these misunderstandings that lead to the spread of trolling.”
These findings offer a fresh perspective on how everyday players experience and react to ambiguous situations. “They should take away that online gameplay can be extremely chaotic and that it is very easy to make mistakes in the midst of this chaos when deciding if a player is being a troll or not,” Velez said.
Players can use this knowledge to reflect on their own habits. “So try to avoid using the signs of trolling I identify in the paper (i.e., auxiliary cues) as quick mental shortcuts for making snap judgments about others,” Velez advised. “But also be aware of how easy it is for other players to perceive your own behaviors as trolling, even when you are not trying to be one.”
The authors suggest that developers could use this information to disrupt the cycle of trolling. If minor, harmless behaviors are given small in-game rewards, targets might be less likely to perceive them as malicious. Additionally, training community moderators to recognize the specific clues that trigger retaliation could help defuse conflicts before they spread.
“The long-term goal is to implement practical steps to stop the spread of trolling through this process,” Velez said. “The paper offers some recommendations but with this approach towards trolling, game developers can devise many interventions tailored to their games and communities.”
A primary limitation of the study is its reliance on prerecorded, staged gameplay. Participants passively observed the video rather than actively playing the game themselves. The cognitive load and emotional highs of active gameplay might alter how players process environmental clues in real-time. The findings might overestimate how deeply players analyze the intentions of others during fast-paced matches.
The experimental setup intentionally excluded voice and text chat to keep the behavior ambiguous. In real gaming environments, players frequently communicate their intentions directly. The presence of verbal communication could easily override the subtle environmental clues examined in this research.
Future research could explore how these interpretations work in other digital spaces. Virtual reality environments and the Metaverse rely heavily on non-verbal actions that could be misinterpreted by users. Scientists also plan to investigate how live streaming shapes gaming norms, as passive viewers might adopt toxic expectations simply by watching others play.
The study, “Trolling is in the Eye of the Beholder: An Attributional Approach to Predicting Perceptions of Trolling and Vigilantism in Online Games,” was authored by John A. Velez, Ines Hyunju Yu, Xan Smith, Sung Hyun Lee, Seung Woo Chae, and Yuxian Lin.
Leave a comment
You must be logged in to post a comment.