Elsevier

Applied Ergonomics

Volume 35, Issue 6, November 2004, Pages 549-556
Applied Ergonomics

Psychophysiological evaluation of simulator sickness evoked by a graphic simulator

https://doi.org/10.1016/j.apergo.2004.06.002Get rights and content

Abstract

The present study investigated the effects of simulator sickness, as an important bias factor on evaluation of emotional changes under the controlled condition of driving a car for 60 min at a constant speed (60 km/h) in a graphic simulator. Simulator sickness was measured and analyzed every 5 min using both subjective evaluation and physiological signals. Results of the subjective evaluation showed there was a significant difference between the rest and the driving conditions 10 min after the main experiment started and that the level of difference increased linearly with time. Analysis of the central and the autonomic nervous systems showed the significant differences in δ, θ, α and β bands of an electroencephalogram (EEG), skin temperature, and the R–R interval between the rest and the driving conditions after about 5 min from the start of driving. In particular, there was the highest correlation between parameter of θ and subjective evaluation, and thus θ was considered an effective physiological parameter for numerically evaluating simulator sickness. The results indicate that physiological changes due to simulator sickness can be a bias factor in evaluation of human sensibility.

Introduction

Human emotion is evoked reactively and immediately to information coming from the external environment, including both temporal and spatial changes. To date, methods for measuring human emotion have included subjective evaluation using questionnaires, as well as objective measurements of physiological signals produced by some external stimuli.

The use of an electroencephalogram (EEG) as a way to measure human emotion or sensibility has recently increased (Elul, 1972, Fox, 1991, Hinrichs and Machleidt, 1992, Kostiunina and Kulikov, 1996). For instances, Davidson and Fox (1982) found that when 10-month-old infants saw video tapes displaying sadness and happiness, their brain waves showed laterality on EEGs. That is, positive emotions such as happiness and pleasure activated the left side of the frontal lobe, whereas negative emotions such as sadness and unpleasantness activated both sides of the parietal lobe equally. Masago et al. (2000) showed a significant change in α band wave at the temporal and the parietal lobes when a pleasant odor was presented. Peter et al. (1995) reported the increases of α and β band waves, in comparison with the rest condition, when the pleasant odor such as phenylethyl alcohol was presented, and an even larger increase was detected when the unpleasant odor of valeric acid was presented.

Relating to the autonomic nervous responses, Elliott (1974) reported that heart rate increased under threat or rage. Ekman et al. (1983) observed the distinctive heart rates for various facial expressions. Levenson et al. (1990) reported differences in the galvanic skin response (GSR) to positive and negative emotions.

Most of these findings were obtained in laboratory environments, since emotion or sensitivity induced in the carefully controlled laboratories could provide the noise-controlled physiological data concerning human emotion. Because many of the results found in the laboratories, however, were obtained in “static conditions” with motionless subjects, they cannot necessarily be generalized to more dynamic conditions involving motions of the participants. As a result, many researchers have become interested in measuring human emotion in more dynamic conditions to attain greater generalizability of findings.

Some of research measuring physiological signals in dynamic conditions were conducted to identify factors that could influence driving performance. Those factors that affect driving performance include illness, sleep deprivation, alcohol or drug intake, and a driver's overload and/or underload (Seppala et al., 1979, Smiley and Brookhuis, 1987, Karel et al., 1993). According to analysis of EEG data, driving performance declined dramatically after prolonged driving (Brookhuis et al., 1985). In particular, an increase of power in α band (8–13 Hz) of the EEG spectrum was associated with a decrement of driving performance.

Heart rate and its variability were also presented as indexes of mental workload (Mulder, 1985, Aasman et al., 1987). Another parameter, relating to driving performance, was the temperature inside the car (Wyon et al., 1989, Wyon, 1996). In addition, research has measured the levels of fatigue while driving and compared the levels of fatigue under different workload conditions in dynamic conditions (McDonald, 1984).

Recent studies have used physiological signals to measure emotional changes in natural environments. For example, one study identified the varied levels of tension due to the different speeds of a car, while another compared the responses induced in different driving conditions, such as a constant speed condition and a sudden-stop-sudden-start condition, by analyzing the activation levels of the autonomic nervous system (Kim et al., 2000, Min et al., 2002). These studies suggest the possibility that we can measure the stable physiological signals even in dynamic environments. In particular, they showed that the level of activation of the autonomic nervous system increased as the level of tension increased. In addition to an increase of the level of the autonomic nervous system, an increase in tension led to the increases in heart rate and skin conductance, lower skin temperature, and the decrease in the photoplethysmogram.

With regard to dynamic conditions, we have to consider which type of dynamic environment should be the basis for research, a natural environment including both vibration and a changing visual scene, or a simulator. If a natural environment (i.e., a real road for driving) is selected, we will have several limitations. First, strictly controlled stimuli cannot be presented. Second, a risk of accidents always exists during doing natural field experiments. Third, it is difficult to analyze data compiled in the field because the physiological signals of EEGs and electromyograms (EMGs) are very sensitive to noise. To overcome these limitations and to study human emotional responses to natural and dynamic environments, studies using simulators similar to the natural environments are carried out.

Although virtual reality systems such as simulators are employed in research where reality is in demand, problems still exist. Many researchers have reported that side effects, referred to as simulator sickness (vomiting, nausea, paleness, cold sweats, languor, confusion, lack of concentration, feeling of crammed brain, blurred vision, and tired eyes) resulted from the use of simulators (Kennedy and Fowlkes, 1992, Kennedy et al., 1993, Kennedy et al., 1997, So and Lo, 1999). After participating in simulator experiments, most participants reported physiological and psychological uneasiness such as dizziness, headache, tired eyes, upset stomach, and intense nausea (Crowley, 1987, Ungs, 1988, Braithwaite and Braithwaite, 1990, Lerman et al., 1993, So, 1994, Regan, 1995).

Furthermore, changes in some physiological parameters, such as decreases in α waves, increases in δ waves, an increase in heart rate, an increase in GSR amplitude, and a decrease in skin temperature, were also reported as a result of prolonged exposure to virtual reality (Cobb et al., 1999).

Simulator sickness seems to have some relationship to personal factors (e.g. age and gender), simulator factors (e.g. lags), factors related to task performance (e.g. degree of control), and the secondary effect of such phenomena as prolonged presence of an afterimage. Consequently, efforts were made to overcome technical problems such as making scenes more realistic and speedy, and to develop statistical tools for extracting more stable results from physiological and subjective responses (Cobb et al., 1999).

A simulator has some important benefits such as presenting reality in a very similar way so that human emotion can be measured under strict controls, as well as eliminating the noise that could arise in real situations. It is necessary, however, to measure simulator sickness objectively and quantitatively for experiments using simulators as substitutes for the natural environment, since simulator sickness can be a bias factor in the physiological signals obtained.

This study was proposed to quantify simulator sickness, as a bias factor, based upon psychophysiological measures. To do this, changes in simulator sickness were observed through measurements of both subjective and physiological (EEG, GSR, ECG, and skin temperature) responses while driving in a fixed-based graphics simulator.

Section snippets

Participants

Participants were 20 healthy adults (10 males and 10 females) whose ages ranged from 20 to 28 years with an average age of 23.4±1.8 years. They were requested not to smoke, drink caffeine, use drugs, or drink alcohol, all of which could influence the central and autonomic nervous system for a week prior to the main experiment.

Apparatus

To process graphic information, a 3D graphic engine based on an Open GVS 4.2 was used. An NEC MT-1030+ LCD projector was used to project images of 30×25inch(h×v) field of

Subjective evaluation of simulator sickness

All participants showed an increase in the level of nausea, discomfort in oculomotor activity, and disorientation with the experimental time, as shown in Fig. 3. That is, they began to feel nausea and disorientation 10 min after the main experiment started and discomfort in oculomotor activity 15 min later. These three responses were significantly different from those of the reference (rest) period. The total simulator sickness score, in which all three scores were added, also increased gradually

Discussion

The study was set up to measure and analyze objectively and quantitatively simulator sickness, recognized as a bias factor intruding on the true evaluation on human emotion in dynamic environments, so that pure emotional changes from the dynamic environment can be extracted from physiological signals. In addition, through insights obtained from the evaluation of simulator sickness, the study intended to provide objective and reliable methods of evaluating simulator sickness.

One of the

Conclusion

In summary, this research looked at changes of subjective stimulator sickness over time; it examined the correlations between responses of the central and autonomic nervous systems and the responses on the SSQ. The θ/total parameter was used to indicate simulator sickness. The physiological changes that resulted from simulator sickness could be a source of bias in objectively evaluating human emotion using physiological signals. In particular, changes in the level of the α/total and β/total

Acknowledgements

This research was supported by a grant of Korea Science & Engineering Foundation (R11-2002-103).

References (36)

  • H. Hinrichs et al.

    Basic emotions reflected in EEG-coherences

    Int. J. Psychophysiol.

    (1992)
  • B.C. Min et al.

    Autonomic responses of young passengers contingent to the speed and driving mode of a vehicle

    Int. J. Ind. Ergon.

    (2002)
  • J. Aasman et al.

    Operator effort and the measurement of heart-rate variability

    Human Factor

    (1987)
  • G.G. Berntson et al.

    Cardiac psychophysiology and autonomic space in humansEmpirical perspectives and conceptual implications

    Psychol. Bull.

    (1993)
  • M.G. Braithwaite et al.

    Simulator sickness in an army simulator

    J. Soc. Occup. Med.

    (1990)
  • K.A. Brookhuis et al.

    The effect of several antidepressants on EEG and performance in a prolonged car driving task

  • S.V.G. Cobb et al.

    Virtual reality-induced symptoms and effects (VRISE)

    Presence

    (1999)
  • J.S. Crowley

    Simulator sicknessa problem for Army aviation

    Aviat. Space Environ. Med.

    (1987)
  • R.J. Davidson et al.

    Asymmetrical brain activity discriminates between positive and negative stimuli in human infants

    Science

    (1982)
  • P. Ekman et al.

    Autonomic nervous system activity distinguishes among emotions

    Science

    (1983)
  • R. Elliott

    The motivational significance of heart rat

  • M.R. Elul

    The genesis of the EEG

    Int. Rev. Neurobiol.

    (1972)
  • N.A. Fox

    If its not left. its right. Electroencephalograph asymmetry and the development of emotion

    Am. Psychol.

    (1991)
  • A. Karel et al.

    The use of psychophysiology to assess driver status

    Ergonomics

    (1993)
  • R.S. Kennedy et al.

    Simulator sickness is polygenic and polysymptomaticImplications for research

    Int. J. Aviat. Psychol.

    (1992)
  • R.S. Kennedy et al.

    Simulator sickness questionnaireAn enhanced method for quantifying simulator sickness

    Int. J. Aviat. Psychol.

    (1993)
  • R.S. Kennedy et al.

    Disorientation and postural ataxia following flight simulation

    Aviat. Space Environ. Med.

    (1997)
  • C.J. Kim et al.

    Study of Autonomic Responses due to the Vehicle Speed Change

  • Cited by (97)

    • Tracking motion sickness in dynamic VR environments with EDA signals

      2024, International Journal of Industrial Ergonomics
    • Effects of unlimited angular motion cue and cue discrepancy on simulator sickness

      2023, Expert Systems with Applications
      Citation Excerpt :

      First, the verbally stated SS severity ratings were compared with the physiological responses to determine their relationships. The physiological data were averaged over two min and compared with the subjective severity captured at the end of the measurement period, as the subjective SS detection had a delay relative to the measured biosignals (Min et al., 2004). Overall, the reported SS severity ratings are negatively correlated with the HR (r(676) = −0.141, p < 0.001) and TEMP (r(676) = −0.175, p < 0.001), whereas the EDA is positively correlated (r(676) = 0.082, p = 0.032).

    • Estimating objective (EEG) and subjective (SSQ) cybersickness in people with susceptibility to motion sickness

      2022, Applied Ergonomics
      Citation Excerpt :

      For example, Chen et al. (2010) used a VR-based driving simulator to induce motion sickness and found an increase in alpha and theta power in the parietal and motor areas. Min et al. (2004) evaluated simulator sickness during car driving and observed that delta power increased in Fz and Cz electrodes. Naqvi et al. (2015) compared 2D and 3D movies to visually induce motion sickness and found that the frontal-theta power decreased over time in 3D conditions.

    View all citing articles on Scopus
    View full text