Effects of Constant and Time-Varying Display Lag on DVP and Cybersickness When Making Head-Movements in Virtual Reality

Abstract

When HMD users move their heads in virtual reality (VR), display lag creates differences between their virtual and physical head pose (DVP). This study examined whether objective estimates of DVP could predict experiences of cybersickness during simulations with three different types of added lag: (1) Constant lag (where the display was always delayed by 250 ms); (2) Predictable time-varying lag (where delays alternated between 0 and 250 ms every 5 s); and (3) Random time-varying lag (where delays alternated between 0 and a randomly determined value, up to 250 ms, every 1–5 s). Constant, Predictable, and Random added lag were found to generate similar levels of cybersickness—with all three conditions producing more severe sickness than the Baseline lag control. Consistent with our DVP hypothesis, the spatial magnitude and temporal dynamics of our participants’ DVP were both found to be reliable predictors of their cybersickness in all display lag conditions tested.

Keywords:

• Head-mounted display
• motion sickness
• cybersickness
• virtual reality
• motion-to-photon latency
• sensory conflict

Data availability statement

Data will be made available upon request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Note: display lag is sometimes also referred to as system latency, or motion to photon latency.

2 Power analysis performed with G*Power using data from a recent display lag/DVP study by Palmisano et al. (2023), indicated that a sample size of 3 should have been able to detect differences in the sickness ratings between the 4 LAG TYPE conditions in the current study (assuming α = 0.05, 1 − β = 0.8 and a Cohen’s f of 1.197). The “makeLmer” and “powerSim” functions from the “simr” library in R were then used to conduct a power analysis for linear mixed models examining the relationships between our DVP predictor variables and cybersickness severity (e.g., Equation 1—see Green & MacLeod, 2016). This power analysis, which involved 500 data simulations, used fixed intercept and slope values, as well as random intercept and residual variance values, for these relationships from Palmisano et al. (2023). The results showed that a sample size of 7 should have been able to detect relationships between the different DVP measures and the sickness ratings in the current study (again assuming α = 0.05 and 1 − β = 0.8). While the above suggests that fewer than 10 participants should have been sufficient for this experiment, there are also well-documented individual differences in susceptibility to motion sickness (see Tian et al., 2022). Thus, our plan for the current study was to test at least as many participants as Palmisano et al. (2023) (i.e., more than 21).

3 Keshavarz et al. (2022) recently concluded that physiological measures (such as EDA, ECG, EGG, respiration, and skin temperature) are “not able to accurately detect and/or predict the onset or severity of [motion sickness]” (p. 23). Thus, we used questionnaires and rating scales to assess sickness severity in this study. The SSQ was used because it is a well-established tool for measuring cybersickness (Rebenitsch & Owen, 2016). We also used a quick rating scale (the FMS) which has been validated against the SSQ (Keshavarz & Hecht, 2011, 2014; Reinhard et al., 2017).

4 PeakDVP contributed to predictions of both FMS ratings and SSQ-T scores when it was the only predictor in this simple model: Sickness severity ∼1 + peakDVP + (1 | Participant) [FMS model information: R²_Marginal = 0.09; R²_Conditional = 0.56; Log Likelihood = −214.71; AIC = 435.51; BIC = 446.52; ICC = 0.52. SSQ-T model information: R²_Marginal = 0.12; R²_Conditional = 0.59; Log Likelihood = −330.02; AIC = 672.75; BIC = 677.14; ICC = 0.54].

5 Machine learning (and deep learning) models are currently popular for studying cybersickness (e.g., Jeong et al., 2023). However, based on the size of our dataset, it was deemed that the use of machine learning models would not be appropriate.

6 Additional model information: R²_Marginal = 0.16; R²_Conditional = 0.63; Log Likelihood = −208.48; AIC = 430.23; BIC = 442.62; ICC = 0.56.

7 Additional model information: R²_Marginal = 0.18; R²_Conditional = 0.65; Log Likelihood = −320.44; AIC = 667.33; BIC = 666.55; ICC = 0.58.

8 Additional model information: R²_Marginal = 0.06; R²_Conditional = 0.46; Log Likelihood = −215.22; AIC = 437.74; BIC = 447.54; ICC = 0.43.

9 Additional model information: R²_Marginal = 0.04; R²_Conditional = 0.67; Log Likelihood = −331.83; AIC = 678.05; BIC = 680.76; ICC = 0.65.

10 Note: These estimated lag values (e.g., ∼254 ms for the Constant condition) were the amount of experimentally added display lag in the condition (i.e., 250 ms in this specific case) plus the average effective system lag (∼4 ms).

11 Except perhaps when there was a brief spike in the Baseline display lag or an abrupt change in the participant’s head-movement.

Additional information

Funding

This research was supported by an Australian Research Council (ARC) Discovery Project (DP210101475).

Notes on contributors

Stephen Palmisano

Stephen Palmisano is a Professor of Psychology at the University of Wollongong. He was awarded his PhD from the University of New South Wales in 1997. His primary research interests are self-motion perception (especially visual illusions of self-motion—known as vection), stereoscopic depth perception, and motion sickness.

Robert S. Allison

Robert S. Allison is a Professor of Electrical Engineering and Computer Science at York University and director of the Centre for Vision Research. He was awarded a PhD from York University in 1998. His research enables effective technology for advanced virtual and augmented reality and the design of stereoscopic displays.

Rodney G. Davies

Rodney G. Davies is a researcher in the Perception and Action Laboratory at the University of Wollongong. He provides programming assistance for its Immersive Virtual Reality experiments and has a strong interest in researching and developing systems for improving end-user experiences in head-mounted VR displays.

Peter Wagner

Peter Wagner is a Research Associate in the Sensory Processes Research Laboratory at the University of New South Wales. His current research aims to create immersive devices for improving “extreme” virtual reality (VR)—where stimulations of visual, acoustic, proprioceptive, and vestibular systems enhance user experiences in head-mounted display VR.

Juno Kim

Juno Kim is an Associate Professor in the School of Optometry and Vision Science (UNSW Sydney). Awarded his PhD in Psychology from the University of Sydney in 2005, his research aims to use virtual reality to understand the complex multisensory processes involved in human movement and perception.