The effect of internal and external fields of view on visually induced motion sickness

Abstract

Field of view (FOV) is said to affect visually induced motion sickness. FOV, however, is characterized by an internal setting used by the graphics generator (iFOV) and an external factor determined by screen size and viewing distance (eFOV). We hypothesized that especially the incongruence between iFOV and eFOV would lead to sickness. To that end we used a computer game environment with different iFOV and eFOV settings, and found the opposite effect. We speculate that the relative large differences between iFOV and eFOV used in this experiment caused the discrepancy, as may be explained by assuming an observer model controlling body motion.

Introduction

With the increasing quality, size, and use of computer generated images, there seems to be a trend in increasing incidence of sickness experienced by people viewing these images, according to reports on several internet forums. Here, we will use the term cybersickness representing sickness caused by viewing dynamic image content as generated by computers in particular. This refers to playing computer games for leisure and for training and simulation (also called serious gaming). In general this is also referred to as visually induced motion sickness, which not only includes cybersickness, but also simulator sickness (typically observed in flight and driving simulators, irrespective of an additional physically moving base) or cinerama sickness (referring to wide screen projection in general). Studies of visually induced motion sickness mainly concern simulator sickness (e.g., Kennedy et al., 1989, Kennedy et al., 1994), and sickness may have a negative effect on the effect of training for several reasons, the most simple being a possible drop-out rate of up to 50% (Reed et al., 2007). The latter also holds for serious gaming in general, as well as testing the affective appraisal using virtual environments (Van der Spek et al., 2007). Visually induced motion sickness, or cybersickness, therefore deserves special attention with the aim of minimising the negative effects thereof.

From a theoretical point of view, cybersickness may be understood by the control of body motion (Oman, 1982, Bos and Bles, 2002, Bos et al., 2008). When moving around on Earth, our central nervous system (CNS) has to control the muscles in the body very accurately. Failing may result in falling. Adequate feedback using accurate sensors is required and both the eyes and the organs of balance play a major role in this respect. However, the visual system is relatively slow¹ and only signals velocity and position. The vestibular system is fast,² but only signals acceleration, including gravity, which is insufficient by itself too. The integration of visual and vestibular signals seems to be optimal for healthy people in natural conditions, i.e., for self-propelled motion. When we are exposed to unnatural motion, however, like being moved by a car, ship, or airplane, or when visual motion is unaccompanied by physical self-motion as in virtual environments, the solutions exploited by the CNS may appear to be less optimal. As in any feedback system, an error signal is probably used by our CNS to move the output (body motion and attitude) towards a desired state. It is assumed that this error signal correlates with motion sickness in general and cybersickness in particular (Bos et al., 2008). The crucial role of the inner ear in this respect is given by the fact that people whose organs of balance do not function, seem not to get sick from motion, even not from visual motion only (e.g., Irwin, 1881, Money, 1970, Cheung et al., 1991; although Johnson et al., 1999 did find some sickness in bilateral labyrinthine defective patients induced by pseudo-Coriolis simulation). Anticipation seems to be crucial as well, as exemplified by the fact that car drivers rarely get sick, while passengers do (Rolnick and Lubow, 1991). Previously, it has been shown that a variety of observations can all be explained by taking into account visual–vestibular interactions within the framework of an observer or internal model, or neural store. This internal model predicts self-motion based on previous experiences to account for sensor imperfections and neural delays (Bles et al., 1998, Bos and Bles, 1998, Bos and Bles, 2002). A conflict between integrated sensory signals and signals from the internal model is assumed to be correlated with motion sickness in particular. Bos et al. (2008) elaborated this model regarding visually induced motion sickness.

One factor contributing to cybersickness concerns the extent of the visual image, or field of view (FOV). There are, however, two distinctly different FOVs at issue. One concerns the FOV value used by the image generator to compute the 2D projection from the virtual 3D scene, a value that will be referred to as the internal field of view (iFOV). The other concerns the FOV subtended by the display from the actual observer's point of view, which will be called the external field of view (eFOV).³ A main effect of the eFOV is already known from studies on balance (e.g., Duh et al., 2001), and sickness and several other factors (e.g., Lin et al., 2002). The general observation is that people get more instable, more sick, and experience more presence with larger eFOVs. Kolasinski (1995) hypothesized that the design viewpoint of the imagery should coincide with the actual viewpoint in order to lower the incidence of cybersickness. The importance thereof was also assumed by Stanney et al. (1998), noting that both small and wide FOVs may induce nausea. Using a head-coupled virtual interface, varying only the iFOV, Draper et al. (2001) did indeed find the least sickness using a scale factor of one (i.e., iFOV = eFOV). The explanation may be that natural conditions are characterized by congruent internal and external FOVs, why the internal model as mentioned above likely assumes this congruency as well. This even holds in the unnatural condition of viewing incongruent FOVs and the sensory signals will then be in conflict with those of the internal model. A discrepancy between the iFOV and eFOV may thus be understood being a cause of cybersickness in addition to the FOVs per se. Different from Draper et al. (2001), who only varied the iFOV, we therefore performed an experiment studying sickness varying both the iFOV and the eFOV.