Unveiling Road Rage Dynamics: Recreating and Modeling Road Rage in Audiovisual and Simulating Environments Based on Real-World Footage

Paper Title

Unveiling Road Rage Dynamics: Recreating and Modeling Road Rage in Audiovisual and Simulating Environments Based on Real-World Footage

Publication Info

• Topic area: Road rage dynamics and digital intervention methods.
• Keywords: Road rage, audiovisual environments, simulated environments, R3-Ftg dataset, anger modeling, parasympathetic rebound, HCI, emotion regulation, dynamic damping function, trigger identification.

Background and Problem

• Problem / challenge: Existing digital interventions for road rage focus on post-response detection and regulation, which can be too late to prevent incidents. Additionally, these methods face generalization challenges due to domain-specific training datasets.
• Significance: Road rage poses significant safety risks, including car crashes, physical confrontations, and fatalities. Understanding its dynamics is critical for developing pre-response interventions to mitigate these risks.
• Motivation and related work: Prior research has explored psychological models of road rage and digital interventions, including measurement, induction, detection, and regulation. However, these studies often rely on artificial scenarios or datasets that lack ecological validity. This paper aims to address these gaps by modeling road rage dynamics using real-world footage and recreating authentic scenarios in controlled environments.

Solution

• Proposed approach: Development of the *Real Road Rage Footage (R3-Ftg) dataset and recreation of anger-inducing scenarios in audiovisual and simulated environments to study road rage dynamics.
• Novelty:
  1. Creation of the first dataset (R3-Ftg) with fine-grained annotations of real-world road rage footage.
  2. Recreation of authentic road rage scenarios in dual environments (audiovisual and simulated) for controlled experimentation.
  3. Modeling road rage dynamics using second-order damped oscillation functions to capture the "slow-rise, fast-decay" phenomenon.
• Procedure and key techniques:
  1. Collection and annotation of 81 real-world road rage clips, labeled with physical/traffic conditions and interaction events.
  2. Selection and recreation of the most anger-inducing scenes in audiovisual and simulated environments.
  3. Recruitment of 52 participants with high trait anger for experiments, recording self-reports and EEG data.
  4. Development of a dynamic damping model to characterize road rage temporal patterns.

Results

• Concrete findings:
  • Road rage was successfully induced in both environments, with simulated environments providing more realistic experiences.
  • A "slow-rise, fast-decay" pattern was observed in road rage dynamics, modeled using asymmetric damping functions.
  • Self-reports and EEG measures showed strong correlations, with peak anger occurring around minute 7 in both environments.
• Advantage over baselines: The simulated environment elicited higher anger intensity than the audiovisual environment, though statistical significance was not achieved after rigorous correction.
• Experiments / evaluation:
  • Participants: 52 individuals with high trait anger (aged 22.26 ± 2.74 years, 14 females, 38 males).
  • Metrics: Self-reported anger levels (Likert scale) and EEG indicators (βH + γ)/(θ + α).
  • Experimental phases: Calm, stimulation, and recovery, with stimuli derived from annotated real-world footage.
• Limitations and future work:
  • Limited participant diversity (mostly young males); future studies will focus on older and female drivers.
  • Exclusion of incidental affect factors like in-cabin conditions.
  • Moderate effect size differences between environments require larger sample sizes for verification.
  • Residual emotional discomfort post-experiment needs further investigation.

Summary

This study addresses critical gaps in road rage research by creating the R3-Ftg dataset and recreating authentic scenarios in controlled environments. Through experiments with 52 participants, it identifies a "slow-rise, fast-decay" pattern in road rage dynamics, modeled using second-order damped oscillations. The findings provide actionable insights for pre-response regulation strategies and pave the way for real-time trigger identification systems. Future work will focus on expanding participant demographics, refining experimental conditions, and addressing ethical concerns related to residual emotional discomfort.