With the Oculus Rift and Project Morpheus somewhere on the horizon for consumers, and the work that Valve is doing, virtual reality (VR) is once again a hot issue for games with all the interest, hype, and business hypothesising that comes along with it.
One often mentioned issue with VR is that some people can feel sick when using it. Indeed, when Sony showed off the Morpheus at GDC this year they warned people if they started to feel sick to let the attendants know and stop playing. In academia this is often referred to as Simulator Sickness (or sometimes “Cybersickness”, if you want to go all Gibsonian).
Simulator sickness is a real problem for some people when using any simulator, although VR is particularly notorious, likely because of the sensory immersion, latency issues, and the added weight of a headset. Simulator sickness is also an issue that is of particular interest to me given my background working with driving simulators. As such, this article aims to lay out what the current science, that I am aware of, has to say about simulation sickness, what it is, why it occurs, and what developers and players can do about it.
Simulation Sickness is a syndrome, which can result in eyestrain, headaches, problems standing up (postural instability), sweating, disorientation, vertigo, loss of colour to the skin, nausea, and - the most famous effect - vomiting. It is similar in effects to motion sickness, although technically a different thing. Simulation sickness can occur during simulator or VR equipment use and can sometimes persist for hours afterwards.
Furthermore, it is possible to use a simulator/VR set without simulation sickness only to get sick once you return to reality, particularly after extended use. Sailors called this “getting your land legs back” in relation to sea sickness, but perhaps we should take the cyberpunk approach and call it “Reality Sickness”. Also, people who use a simulator or VR for extended periods of time have been reported by some to have simulation sickness “flashbacks” later. Both of the above factors led the US military, according to some reports, to have a 24 hour ban on flight for any pilots after using a simulator.
One of the big issues with simulation sickness is that it is understudied (and some studies that exist are old, have questionable methodology, or rely on anecdotes), and that there is no universally agreed explanation for why it occurs. As such, the aforementioned ban by the US military may be an over-reaction or it may not.
There is no firm agreement on why, but there are a few theories. I will go briefly into them below, but if you are not interested in theorising, and just want to read tips on how to reduce simulator sickness, please do skip to the next section:
Cue Conflict Theory
This is probably the most dominant theory and simply put says that simulation sickness is caused by a mismatch between what we expect to see and feel and what we are actually seeing and feeling. Basically, this theory says that our body has an expected sensory input (based on past experiences and biology) and if that sensory input is different, for example you are running around in a VR first person game but in real life are sitting on your couch, we can experience simulation sickness.
The supporting evidence for this theory is:
- People who are more experienced at an activity in reality are more likely to experience simulation sickness than novices. According to this theory this is because experts in the activity have a stronger/more sensitive sensory expectation and therefore virtual reality is more likely to provide the wrong input.
- That adding motion to a simulation can help. However, it must be accurate otherwise it just adds even more to the sensory conflict. So, there is still an argument in the scientific community about the merit of adding movement to simulators in terms of a cost-benefit trade off.
- That extended exposure can reduce symptoms. The argument being that you form new expectations related to the simulation - this then also explains possible sickness upon returning to reality where your expectations have changed.
Evidence against this theory:
- It is a theory that has low reliable predictive value. In that it cannot be used to reliably predict who will or will not get motion sickness. Basically, people don’t get sick every time they experience different sensory input than they expect, just sometimes.
- The theory doesn’t explain why sensory expectations not matching input would cause such an extreme reaction. Throwing up and falling over are not usually very evolutionary useful reactions, so why feel that way when the only issue is that your eyes say you are moving but the rest of your body says you are not?
Postural Instability Theory
The next theory is focused on the falling down and feeling dizzy part of simulation sickness and is promoted as an ecological theory that doesn’t rely on fluffy cognitive things like “expectations”. This theory says that our body is constantly making small adjustments to not fall over and relies on accurate sensory input to do so. When the sensory input is off the small adjustments also get thrown off resulting in simulator sickness.
Evidence for this theory:
- Balance problems and dizziness can precede other simulation sickness symptoms.
- Helps explain why simulation sickness goes away over time, basically your body adjusts. It also explains why people who are more experienced with a real life activity are more likely to get simulator sick (they have more to “train away”).
- Postural instability doesn’t always precede simulation sickness.
- It, like cue conflict theory, has low predictive value.
- Why does this lead to such an extreme “blllarrrhggghh” response?
This theory mostly exists because of the last problem with conflict theory and postural instability theory in that the reaction to simulation sickness seems extreme. Poison Theory says that, evolutionarily speaking, one of the only times you will experience sensory input that is different than what you expect, or when you get all dizzy and have problems standing up, is when you have been poisoned. So what is evolutions best trick to get rid of poison? Be sick.
Evidence for this theory:
- People get sick and fall down…
Evidence against this theory:
- Low predictive value (notice a theme?). It is an evolutionary story that sounds good, but can’t really be reliably and accurately used to predict who will get sick under what conditions.
- Why are young children resistant to simulation sickness? They are more vulnerable to toxins, so shouldn’t this defence be stronger?
- Doesn’t stand by itself very well, so it is sometimes referenced just as an explanation for the pukey parts of the other two theories.
Other, lesser known, theories include:
- Eye movement theory: Suggests that certain stimuli cause eye movements that create tension in the eye muscles which stimulate the vagus nerve leading to sickness.
- Subjective vertical mismatch theory: An alteration of conflict theory that suggests the most important factor is the mismatch between a users sense of verticality vs the sensor input they are receiving about their verticality from the simulation.
- Negative Reinforcement Model: Like Poison theory, this is an evolutionary theory that argues that movement that would generate sickness, or mismatched sensory information, suggests weakness or doing something that could make you weak. Therefore, to stop you doing it you get sick.
So, there is no agreement on exactly what causes simulation sickness, and indeed the theories overlap quite a bit, so the likely answer is a mixture of these ideas (plus probably some new ones). But despite not understanding what causes it there are some techniques that research suggests developers can do in games that may reduce its occurrence. Please note, these are only guidelines and may have varying levels of influence depending on the game, the hardware, and the person playing.
- Maintain high frame rates and low latency: It isn’t just for game quality reasons that the folks at Oculus, Sony, and Valve have been stressing the need for low latency, responsiveness, and high frame rates. The low persistence tech Oculus is building into the Rift will help, but you still have to pull your weight. Basically, if there is lag between a player input and game output then the chance of simulation sickness increases. So, when it comes to VR if there is choice between better graphics vs low latency, the latency should win out. In terms of strict guidelines, it is hard to say. Some academics cite any latency above 46ms could start to get problematic, others cite 20ms, and indeed, Oculus targets under 20ms. Similarly, avoid motion blurring, which increases the sensation of movement.
A comparison of the low persistence offered by the new Oculus dev kit versus full persistence. Full persistence is much more likely to make you sick.
- Avoid flicker: Flicker and flashing should be avoided, particularly near the edges of vision, where it can be perceived by the brain as a signal for movement.
- Make use of appropriate movement: Basically, match sensory expectations. So if the player is a human character, don’t use unrealistic movement (e.g. strafing) and tie camera movement to the head (i.e. it works like a turret, or… a head…). This includes taking advantage of head tracking when offered. Particularly inappropriate movement that should be avoided includes rapid tilting, rolling, and sinusoidal movement (e.g. up and down, or left and right, wave like movement, particularly at frequencies between 0.05 and 0.8 Hz). One example of these types of movement in games could be certain types of head bob or going up and down stairs - both of which can be troublesome in VR. Another example could be gun sway that fills a significant amount of the player’s vision and could create the illusion of left to right swaying motion.
An example of headbob that could be problematic in VR (This GIF was generated from this video, which is simply a engine demo and is not showing a VR game)
- Limit uncontrolled movement: Related to above, basically if movement is happening without player control, particularly tilting, spinning, or flipping movement, then you are on your way to a vomit comet simulator. This idea of the advantages of being in control is similar to how drivers are less likely to get motion sick than passengers. Examples of uncontrolled movement could be as simple as a death animation where the character ragdolls or falls to the ground, tilting and shaking the camera, or a zoom transition between scenes.
- Think about your camera height: The closer a camera is to the ground, the more the ground fills a player's vision and the faster what they are seeing on the ground will change. This increases the sensation of movement.
A proximity flying wingsuit simulator may be out of the question for those who are vulnerable to simulation sickness. Although, the folks who made Aaaaaaaaa! for the Rift, a base jumping game where surfaces do wizz by, say their game has overcome simulation sickness problems.
- Limit movement through a scene: This is not to say remove control (see above) but rather use a fixed position where the player just has turret-like control of the camera. This of course limits game design, so it is a trade off.
- Limit rapid changes in acceleration (including deceleration) and rapid shifts of perspective: This speaks for itself. But interestingly this also has an application when zooming vision, say with a scope on a gun in a game. Not only could it be a bad idea to have a majority of the screen rapidly zooming in or out, but being zoomed in can also increase the sensory input discrepancy. This is because when you are zoomed in your cone of vision moves faster than expected. To try this out yourself look through binoculars and then try moving around, it can get sickening quickly. Also avoid to sudden death transitions or rewinding/fast forwarding time.
- Get daily news, dev blogs, and stories from Game Developer straight to your inboxSubscribe