Virtual Reality and Human–Machine Interaction in Road Aggression Research: A Systematic Review and Bibliometric Analysis

Road aggression remains an important road-safety concern and should be distinguished from risky driving, as risky behaviour does not necessarily involve hostile or aggressive intent. This study examined the development, methodological characteristics, and intellectual structure of research on virtual environments and human–machine interaction in assessing, explaining, and reducing aggressive driving. We conducted a systematic literature review and bibliometric analysis using Web of Science Core Collection, Scopus, and APA PsycInfo. The searches identified 134 records; after removing 46 duplicates, 88 unique records were screened. Forty-four reports were sought for full-text assessment, four could not be retrieved, 40 were assessed for eligibility, and three were excluded. The final review included 37 studies. Virtual and simulated driving environments were treated as an overarching analytical category encompassing immersive virtual reality, conventional driving simulators, computer-generated driving environments, and relevant human-machine or human-vehicle interaction applications. Technological modalities were considered separately during synthesis. The literature covered anger- and frustration-provoking situations, congestion, impeded progress, interactions with other road users, provocative behaviour, automated vehicle driving styles, and human-automated vehicle interactions. Increased speed, rapid acceleration, reduced safety gaps, risky overtaking, tailgating, and shorter time-to-collision were treated primarily as indicators of risky, conflict-related, or operational driving behaviour and were interpreted as indicative of aggressive driving only when supported by contextual, behavioural, emotional, intentional, or psychometric evidence. Some studies further found that anger or trait aggression did not necessarily predict risky driving behaviour. Methodological limitations included small and insufficiently diverse samples, inconsistent definitions of aggressive driving, heterogeneous scenarios, variation in simulator fidelity, and limited standardisation of behavioural and physiological measures. Future research should integrate validated psychological measures with behavioural, psychophysiological, and vehicle-dynamics indicators to strengthen construct validity and support adaptive training, targeted interventions, and safer human-vehicle interaction systems.