Individual susceptibility varies widely

What it is

The same demo: one person is fine at thirty minutes, another is pale and asking to stop at ninety seconds. That is not “the content is OK for the average user.” Cybersickness susceptibility spans a very wide range across people — from almost never sick to barely able to finish onboarding. A mean, and “we tested twelve people and only two were uncomfortable,” cannot cover that distribution. The product faces the tail, not the mean.

Susceptibility is a trait, not merely whether breakfast happened this time. History with cars, boats, and coasters often predicts who stops first better than this session’s hardware tier.

Why it happens

Tolerance for vestibular–visual mismatch already has large individual variance: canal and otolith gain, speed of habituation, comorbidity of migraine and vestibular pathways, all change how large a mismatch must be before the protective response fires. Instruments such as the MSSQ ask about conflict already encountered in life (cars, boats, attractions). They predict headset response because the underlying mismatch alarm is the same, not because a car looks like a headset.

The distribution is skewed. Most people sit in the middle, a small group is almost immune, a small group is extremely sensitive. Twelve colleagues drawn at random easily yield a sample with no sensitive people, and the conclusion “our experience does not sicken.” Sensitive people are not fussy; their threshold is different. Sex differences appear repeatedly in the literature and are repeatedly challenged on method (sample, reporting style, prior exposure). The operable conclusion is not “women will always be sicker”; it is “do not sign off for everyone on a convenience sample of one sex and one age.”

Studying it

Stratify susceptibility first with the MSSQ or a similar history questionnaire, then run the same stimulus through low, mid, and high groups. Dependent measures remain SSQ/VRSQ and abort time, but results are reported by stratum, not as a grand mean only.

Independent variables: susceptibility stratum, (if habituation is in scope) prior VR hours. Dependent variables: time of first discomfort, abort rate, symptom peak.

Recruitment has to include high-susceptibility people on purpose. Lab ads that say “VR experience” already bias toward people who resist sickness, and the tail is truncated. Writing “people who did not get sick liked this feature” as a result for all users is extrapolating from a truncated sample.

Where it stops holding

State stacks on trait: sleep debt, alcohol, hormonal cycle, anxiety push the same person’s threshold toward sensitive, and it can move within a day. That does not mean trait is absent; it means a single measurement is noisy and needs repeats, or at least a record of state. A never-exposed person’s first response mixes novelty and is not a clean trait reading; the stretch that stays stable after habituation is closer to susceptibility. Vestibular disease and migraine are clinical subgroups; their data are not a rejection region for a general product, but a product that claims “usable by everyone” has to test those subgroups rather than substituting healthy students. Susceptibility predicts risk, not fate: high-susceptibility people can finish in very low-conflict content; low-susceptibility people still sicken on extreme stimuli.

Applying it

• Write high-susceptibility people into the acceptance roster (screened with a history scale, not with “they look like they might get sick”). Playtests of colleagues and enthusiasts only cannot be a susceptibility conclusion.
• Design the default experience so high-susceptibility people can finish onboarding; put stronger visual motion in optional or advanced, not in the first three minutes.
• Provide a clear abort and an intensity switch, and test on high-susceptibility people whether they will actually use them — a switch buried in a deep menu is no switch.
• How to check: split at least two strata on the MSSQ and run the same onboarding. A still-low abort rate in the high stratum is what it takes for that onboarding to hold across the distribution. If only the low stratum can finish, the product’s “does not sicken” is a truncated sample. Mark the high stratum’s abort times on the timeline; those are the places the content has to change.

Related

• Same group: N1.06.1 Conflict between visual motion and vestibular input is the main cause · N1.06.3 Symptoms are delayed and cumulative
• Nearby: N1.13 Causes of Cybersickness and Individual Differences · N4.02 Continuous Locomotion
• Search terms: cybersickness susceptibility · MSSQ · individual differences