Acceleration and turning are the strongest sources of discomfort

What it is

On the same smooth locomotion, the worst seconds are usually not constant forward speed but launch, braking, and turning. Acceleration and turning feed the vestibular system the signals it is most sensitive to — linear acceleration and yaw velocity — while vision reports them and the body does not. Constant-speed straight motion keeps the conflict at “speed without acceleration”; speed changes and turns push it to a peak.

This is which slice of the motion profile sickens, not whether continuous locomotion should exist.

Why it happens

Otoliths respond to linear acceleration; canals respond to angular acceleration. Visual launch and braking are moments the otoliths should have moved and did not; visual yaw is a moment the canals should have moved and did not. Both are transients, and the canals are especially tuned to yaw. Steady optic flow at constant speed is also a lie, but the lie is “there is velocity,” which otoliths do not encode; that conflict is steady-state and partly habitable.

Turning is harsher one layer down: rotational flow sweeps almost the entire retina, a larger area than the radial flow of forward translation. Stack translation and yaw on the same stick input and otoliths and canals are both left hanging; the peaks add. That is why “walk and twist the view with the stick” throws people harder than “stop, then turn.”

Studying it

Split continuous locomotion into a profile rather than one “smooth” level. Typical comparisons on one route: a stick curve with obvious accel/brake versus near-constant speed; head-turn only versus extra stick yaw; translation coupled to rotation versus decoupled.

Sample FMS immediately after each acceleration segment and each turn, not at the end of the whole run. With aligned timestamps, nausea peaks can be locked to specific accel events rather than to “used continuous locomotion.” An SSQ total averages those events away.

Where it stops holding

On a passive vehicle — a roller coaster, a cinematic camera — accel and turning are imposed by the content, and the buffer of “I initiated this” is gone; peaks run higher than the same profile on a stick. A treadmill or a real body turn that already supplies matching acceleration makes the visual accel no longer fake. At very low speeds the accel slice may sit below notice and profile differences collapse. If turning is implemented as a discrete angle step, continuous angular velocity is deleted and the claim about turning-as-a-continuous-profile no longer applies.

Applying it

• Keep stick acceleration and braking short and soft. Avoid jumping from rest to top speed; top speed itself is often not the peak source.
• Decouple translation from stick yaw: turn with the body or head while walking; let the stick do forward/back.
• For large heading changes, drop translation near zero first. Do not stack an in-place twist on full speed.
• How to check: log input on one route. Align FMS peaks to accel and turn segments. If peaks sit on launch, brake, or stick yaw while constant-speed straights stay quiet, change the profile rather than replacing the whole technique with teleport.

Related

• Same group: N4.02.1 Continuous locomotion keeps spatial sense but readily causes discomfort · N4.02.3 Constant-speed straight motion is less uncomfortable than speed change
• Nearby: N4.07 Continuous Locomotion and Discomfort · N4.03 Cybersickness Mitigation
• Search terms: acceleration-induced cybersickness · rotational vection · locomotion motion profile