Detecting_Betrayers_Online-Chaima
#notesFromPaper Authors: Chaima, Magy, Paola Tags: deception
- There exists research on detecting betrayal using anomaly detection
- There are a few assumptions about betrayers, for example, that they have thought processes vastly different from those who do not betray. They expect that individuals who have betrayed will feel guilty, anxious, trapped, and distant from the group.
- similarly, less identification and trust with the team
Related Work
- honeypots have been used to lure attackers
- deceivers tend to be more submissive when trying to evade detection, while being confident and dominant when trying to persuade about innocence.
The game
- Players had to use clues to figure out a stranger’s identity (similar to guess who)
- Players could pass info to the enemy team for an extra $2
- The team was told that someone had been offered the ability to betray (and that they had declined or accepted)
- Used “Affective Indicators” which were stimuli intended to make the participants feel differently about the team
Conclusions and results
- The biggest indicator of potential betrayal was AI-Disengage-B, which is how many text chats were sent after the deception had occurred.
