Inducing and measuring acute stress in virtual reality: Evaluation of canonical physiological stress markers and measuring methods

Abstract

Besides entertainment, Virtual Reality (VR) is becoming increasingly important for professional applications. VR offers a wide range of use, from simulating and dealing with critical or dangerous situations or procedures, competence-oriented learning and testing in an educational context, diagnostic or therapeutic applications in the medical field, or even for tele manipulators and robotics. In most of the mentioned scenarios, measuring the alertness, mental load, stress or relaxation state of the user would be a clear enhancement. That information can serve as a valuable input to control or influence the VR-setting (e.g. for biofeedback) or lead to additional information, for instance as an countable outcome of a simulation or for objectifiable results that can be used for debriefing or diagnostics. Since stress mirrors a neurophysiological response, it is possible to measure the reaction of the human body to a stressor. This leads to the activation of the autonomous nervous system that controls many different physical reactions. To find the best parameter measuring the timing of a sympathetic activation to a predefined stressor (especially with regards to our research of detecting substance craving precisely after a virtual exposure for a VR-based therapy system in addiction medicine), electroencephalography (EEG), heart rate variability (HRV), skin conductance response (SCR), and peripheral skin temperature (ST) were compared against each other regarding a real-time detection of stress among criteria such as the costs for building a wireless sensor, the ease of use in a professional, medical or home use and the robustness against artifacts and the precision of the measurement. We used a 180-degree horror movie as a stressor for a healthy study population and took care about possible confounders, that may influence the subject's stress level or the measurements. All measurement procedures and the used devices complied with medical standards. It was possible to provoke a measurable physiological stress reaction in all subjects through VR, although not all participants subjectively felt scared or stressed. EEG, SCR and peripheral skin temperature turned out to be equally well suited for real-time detection of stress. HRV failed to reliably show the stress response, which does not disqualify this measuring method, but might rather be connected with measurement timeframes and data preparation in our study. When it comes to the ease of use, the robustness of the measurement, costs and the possibility to implement the measuring technique within a wireless system (e.g., finger clips or even the VR hand controllers), we consider the skin conductance response (SCR) as the most promising physiological parameter to objectify stress in a virtual environment.