A new study in Scientific Reports, Nature’s open-access journal, finds that how far a user teleports in virtual reality shapes how well they learn a virtual environment — and that adding a brief optical-flow transition to teleportation made no clear difference in the wayfinding measures. In a 60-participant experiment, teleport distances of 5, 15, and 25 meters produced a measurable trade-off: 25-meter jumps cut travel time but degraded participants' sketch maps of the space, while shorter hops slowed travel but supported environmental learning.
Longer jumps, worse mental maps
The experiment used a mixed design with two teleportation types — no optical flow versus a brief 0.3-second optical-flow transition — crossed with three maximum distances: 5, 15, and 25 meters. Participants first explored routes in a virtual urban environment, then had to find their way back without guidance.
The results split cleanly along two axes. Sketch-map scores — a standard measure of how well someone has internalized a spatial layout — dropped significantly at 25 meters compared with 5 and 15 meters. Travel time, meanwhile, ran longer at 5 meters than at 25, and the mean time per teleportation rose steadily across the 5-, 15-, and 25-meter conditions. In other words, long jumps get you there faster but leave you with a fuzzier picture of where you have been.
What the optical-flow manipulation did not do is just as notable. The study detected no significant between-type differences and no type × distance interactions on either sketch-map or travel-time measures. The 0.3-second flow burst, fixed across all distances, coupled displacement with translation speed — a design confound the authors flag directly, meaning the transition’s role in this trade-off remains specific to the tested configuration rather than a general law.
Simulator sickness stayed low — and one analysis says the flow effect isn’t real
Simulator sickness, the perennial bugbear of VR locomotion research, registered low scores across every condition. That is consistent with teleportation’s core appeal: instantaneous displacement omits continuous viewpoint motion, which is why it dodges the vestibular conflict that makes smooth artificial walking nauseating for many users.
But a methodological wrinkle deserves attention. When the team applied aligned rank transform (ART) analysis to the sickness data, it indicated effects of teleportation type, distance, and their interaction. Those effects were not consistently reproduced by other statistical approaches, and the authors conclude there is insufficient evidence for a stable simulator-sickness effect. For anyone building locomotion systems on the strength of a single statistical method, that is a cautionary data point.
Teleportation cuts learning, sickness stays low
- 25 m teleportation = fastest traversal + weakest spatial learning in this setup
- Brief optical-flow transitions showed no measurable wayfinding benefit or cost
- Simulator sickness stayed low everywhere; the ART-flagged sickness effect failed replication checks
- Findings are scoped to these exact configs: fixed 0.3 s transition, fixed scene variants per distance
What participants actually said
Semi-structured interviews added texture the numbers miss. Shorter-distance teleportation, participants reported, made it easier to observe the environment — you land, you look around, you build a map. Longer jumps felt easier to operate. That subjective split lines up with the objective divergence between travel time and sketch-map scores, and it suggests developers face a genuine UX fork: comfort of operation versus how well users come to know the world you built.
Where this lands in the locomotion toolchain
A museum tour app that wants users to memorize a gallery layout has different optimal parameters than a fast-paced game where traversal speed is king.
The caveat matters: the researchers themselves note the trade-off “remains specific to the tested configurations” because distance conditions were paired with fixed scene variants and a fixed transitions duration that conflated jump length with translation speed. A 25-meter hop completed in the same 0.3 seconds as a 5-meter hop implies very different perceived velocities — a variable no current production engine amortizes automatically.
What the study leaves open
Three questions remain unresolved. Whether variable-length optical-flow transitions — where duration scales with displacement — would change the sketch-map penalty at long distances. Whether other statistical framings would ever validate the transient sickness signal ART surfaced. And how these effects interact with egocentric versus allocentric wayfinding strategies, which the sketch-map instrument measures only indirectly.
For now, the actionable takeaway for interaction designers is a measured one: pick your teleport ceiling based on whether your application rewards speed of traversal or depth of environmental learning — and, on current evidence, a brief flow transition will not buy you a way out of that trade-off.
