UC Berkeley researchers use virtual reality and earthquake simulator to study human risk perception during seismic events

The motivation for the study stems from a recognized gap between knowledge of earthquake safety protocols and actual human behavior during seismic events. Ceferino's prior fieldwork on the 2023 Türkiye and Syria earthquakes found that nearly 40% of building occupants surveyed reported feeling "incapacitated or frozen" due to panic during shaking, while very few successfully executed standard safety protocols such as drop, cover, and hold on. Other documented social responses during earthquakes include running because others were running and rushing to protect family members, behaviors that Ceferino notes can place people in greater danger but are not currently factored into earthquake safety policies or guidelines.

Existing data on perceived risk is limited to post-event surveys, which rely on participants' memories and cannot be reliably linked to measurable physical parameters, such as shaking intensity at a specific location and time, without comprehensive sensor coverage in place. The current study aims to address this limitation by capturing real-time behavioral and perceptual data during controlled simulations. The researchers intend to build a framework for understanding how risk perception — defined as the fear of being injured — is modulated by earthquake shaking intensity.

To create a realistic experimental environment, the research team synchronized the motion of the Shake House, an earthquake simulator located at the California Academy of Sciences, with a purpose-built virtual reality environment depicting objects moving and falling. The team measured the accelerations used by the Shake House to simulate two historical earthquakes — the 1906 San Francisco earthquake, recorded at magnitude 7.9, and the 1989 Loma Prieta earthquake at magnitude 6.9 — and calibrated the virtual reality environment to match those physical parameters. The result was described by Ceferino as a "4D simulation," in which participants experienced a three-dimensional virtual space while simultaneously feeling the physical shaking evolve over time.

Fifty participants took part in the experiment. Prior to entering the Shake House, they completed a survey about their earthquake knowledge and prior experience with large earthquakes. Inside the simulator, they wore VR headsets that placed them in a virtual living room inside a one-story building, with a partially visible kitchen and bedroom. During the simulations, the headsets recorded participants' gaze, allowing researchers to determine what individuals were looking at throughout the shaking sequences. Participants communicated their experience in real time using a VR controller, pressing a button whenever they felt they could be at risk of injury.

Preliminary findings indicated that participants' risk perception responses were not linear. According to Ceferino, risk perception increased dramatically at certain points, particularly when objects began falling and the shaking intensity became sufficient for participants to feel physically off balance. Following the simulations, participants completed a post-experiment survey documenting their perceived injury risk levels, their behavioral responses during the shaking, and additional details about their experience.

The research team's stated objective is to quantify human behavioral responses to earthquakes and identify the key factors that influence decision-making during seismic events. The knowledge gained is intended to inform public education and practical safety guidance, including recommendations about safer movement paths within buildings and the identification of specific hazards such as broken glass or fallen objects. The project is partially supported by the National Science Foundation under grant CMMI N 2410291.