Virtual Reality Aggression Assessment with Social Interaction: Early Evidence for Validity from Two Pilot Studies

Fedde Sappelli^1*Jill Lobbestael²Ditte van Haalen¹Irina Böckmann¹Erik Bulten¹Robbert-Jan Verkes^1,3

• ¹Pompestichting Forensic Psychiatry, Nijmegen, Netherlands
• ²Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands, Netherlands
• ³Radboud University Medical Centre, Nijmegen, Gelderland, Netherlands

Assessment of aggression proneness is important for clinical practice and research. Virtual Reality (VR) is a promising technology for aggression assessment because of the possibility to create scenarios with ecological and external validity, and experimental control; potentially overcoming limitations of traditional methods like self-report trait questionnaires, observation surveys, and laboratory paradigms.This explorative study investigated a VR scenario in a pilot with aggressive patients (n=12) and a followup with students (n=12). The VR scenario consisted of social interactive roleplays with virtual characters (avatars). It consisted of a neutral scene, an instruction scene and two provocative scenes: one with a socially and verbally aggressive, uncooperative female avatar (provocation 1), and one with a dominant, unreasonable, intimidating male avatar (provocation 2). The primary outcome was VR-displayed aggression, assessed with a VR-customized version of the Social Aggression and Dysfunction Questionnaire. Life-time aggression self-report scales were also administered.The main analysis revealed that both patients and students showed higher levels of aggression in provocative scenes compared to neutral and instructional scenes. Exploratory analyses suggested that patients displayed increased aggression in the second provocation compared to the first, while no such difference was observed among students. Comparatively, patients showed more aggression than students in provocation 2, but not in provocation 1. Positive moderate correlations were found between VRdisplayed aggression and trait questionnaires. The current study shows that aggressive behavior can be evoked with our VR aggression assessment scenario, and that the level of aggression can be systematically assessed using a standardized aggression observation scale. Explorative results also imply that the VR scenario has construct, known-group and concurrent validity. The results also imply that VR is potentially closing the correlational gap between behavioral tasks and trait questionnaires. However, the explorative nature of the current study warrants replication. Finally, we outline some scenario aspects that can be further improved, including better physical engagement and standardization.