When the boundary is invisible, people stop moving
Aliases: guardian fade · lost play area · boundary uncertainty
What it is
The grid dies, the room lights are off, and feet pin in place even though the virtual corridor still runs forward. Under play-area uncertainty, locomotion is frozen on purpose: not because a wall was hit, but because the next step might hit one. An invisible boundary pins people harder than a badly drawn one.
This is not the same claim as “cues must appear early.” Early appearance governs approach to an edge. This leaf governs whether people still dare to walk after the edge has left perception.
Why it happens
Once the headset shuts out the real room, position in that room is held up by two things: the grid just seen, and a bodily estimate of “how many steps remain.” When the grid vanishes, the estimate starts to drift. Drift is usually conservative — walk less, rather than probe furniture with a shin. The result is a distinctive stop: the virtual target still ahead, feet already stopped inside the play area, a large unused patch of empty floor left behind.
Uncertainty also changes gait. Stride shortens, weight shifts back, a hand goes out to probe, as in a dark room. Content that reads that stop as “the user does not want to continue” will mis-bind the interaction. Passthrough that pops open and shut, a flickering grid, a redrawn play area, all zero the estimate and freeze more often than having no boundary system at all — because people had been relying on it, and have no spare room model when it goes.
Studying it
Have people walk to the same virtual target with a visible grid, with the grid suddenly gone, and with no play area ever calibrated. Measure arrival, stopping margin from the real edge, and stride change. The load-bearing observation is an interior freeze: stopping far from the real wall. Collision counts may even fall, because people have stopped walking; using collision rate will misread “dare not move” as “safer.”
Ask “where do you think the wall is” and have them point; compare to the real wall to see how far the estimate drifted. Larger drift, earlier freeze.
Where it stops holding
Under optical see-through the real wall stays visible, so this freeze weakens; limits come from social space and furniture, not estimate drift. People who know the room well enough to walk back to the table with eyes shut are less hurt by a missing grid. Seated use barely depends on a floor boundary; the freeze moves to the hands: unseen desk edges pull the hands back. A grid that stays faintly visible at all times reduces freezing, at the cost of presence and of people hugging the edge. In a lab with no boundary system, where people know from the start they are walking blind, behaviour is slow probing, not necessarily pinning — pinning shows up more in the drop from “the grid was just there” to “now it is not.”
Applying it
- Do not fully extinguish the grid while people still need to walk. Far from the edge it can be very faint, but it must stay retrievable, or one-button summonable.
- On tracking loss, play-area redraw, or passthrough toggle, say plainly that the boundary is temporarily unavailable, and pause tasks that need a step. Do not pretend the world is still safe.
- While the boundary is down, send far targets through teleport or stick. Do not keep urging a walk.
- How to check: mid-task, with a spotter present, kill the grid on a must-walk segment. If people stop far from the wall or switch to probing with a hand, the freeze has already happened. Restore a summonable boundary or change locomotion; do not add a louder “keep going” prompt.