Cybersickness follows sensory conflict theory (Reason & Brand, 1975): the brain expects vision and the vestibular system (inner-ear semicircular canals + otoliths) to report self-motion in agreement. In a seated VR headset your eyes report large, fast optic flow — vection — while your vestibular system correctly reports that you never actually moved. That unresolved mismatch is what the nervous system reads as a poisoning cue, triggering nausea.
C(t) = w_fov·fov_n + w_speed·v_n + w_turn·ω_n·(1 − 0.6·snap) + w_lat·L_n
S(t) = clamp( S(t−dt) + k·C(t)·dt − d·dt , 0, 100 )
where fov_n, v_n, ω_n, L_n are the field-of-view, locomotion speed, turn rate and latency each normalised to their slider range, C(t) is the instantaneous conflict index (0–100), and S(t) is the slowly-integrating, slowly-decaying exposure score used for the risk category — mirroring how real Simulator Sickness Questionnaire scores build up over a session rather than reacting instantly.
- Field of view & speed — a wider FOV and faster optic flow recruit more peripheral retina, which drives vection strength and conflict fastest (the single biggest predictor in cybersickness research).
- Turn rate / locomotion mode — continuous smooth turning keeps visual rotation running with no matching vestibular signal; snap-turning (discrete jumps) trades a brief disorientation spike for long stretches with near-zero rotational conflict, which is why most shipped VR titles default to it.
- Latency — the delay between a motion command and the rendered response (tracked here by literally rendering the tunnel from a buffered, time-delayed camera state) adds a prediction-error term found in every major cybersickness study; above ~50 ms it becomes noticeable, above ~150 ms it dominates.
- The small inset shows the two conflicting signals directly: the spinning arrow is the visual heading your eyes see; the ball stays pinned to true-down because the vestibular system correctly senses the headset wearer never rotated.