In-range, locked, and activated need three distinguishable feedback forms

What it is

Mid-air pointing usually passes through three stages: in-range (the hand is seen, the object not yet grabbed), locked (the object is held, not yet committed), activated (the command has landed). Buxton's three-state model writes graphical input as out-of-range / tracking / dragging; mid-air has no physical mouse-down, so the three stages are held open entirely by feedback. The three states need three distinguishable forms, so a user can answer “which step am I on.” Missing contact is why artificial signals are needed at all; what is required here is that the three signals be recognizable from each other.

Why it happens

In-range means “I can be selected”: a cursor appears, the object lifts slightly. Locked means “I am holding it”: the object sticks to the cursor, the outline becomes a grab style. Activated means “this one is done”: a settling change—the object seats, a list pages, a check appears. If all three share one highlight, users cannot know whether releasing now cancels, continues a drag, or submits. Mid-air is especially brittle because no key travel tells the finger “already down.” The form difference can come from shape, motion, timbre, or a haptic pulse, as long as each pair is discriminable when they can be seen together. The three states are progress inside a session, not whether capture is on, and not whether the recognition score is high.

Studying it

Without teaching copy, have participants force-choice the current state from feedback alone. Materials: brightness-only, shape change, motion change, cross-channel combinations. Dependent measures are three-state classification accuracy, hesitation time, and accidental submits on release. Include a control where in-range and locked are deliberately given the same look, to see where discriminability falls. The task must include mid-course abort, so you can see whether locked is treated as already activated.

Where it stops holding

A product with only instantaneous taps and no held grab may crush the middle locked state to a single frame; the three states collapse toward two, and “aim” versus “click” still have to split. In a continuous scrub, “activated” is ongoing writing rather than one settlement, so the third form should be continuous rather than a checkmark. Color-only splits collapse for color-vision difference. In a headset the hand is already in view, so proprioception can partly stand in for in-range; locked and activated still need to be explicit.

Applying it

• Assign each of the three states a form that does not share shape or channel. Write a contrast table. Forbid all three as one ring of light at different opacities.
• Release in locked should cancel or drop; in activated it should already have finished. Mark on a frame-by-frame recording which state the release landed in.
• At acceptance ask “what happens if I let go now.” No answer or a wrong answer means the three states have not been split.

Related

• Same group: C4.26.2 Using one visual cue for all three states leaves users unable to judge progress · C4.26.3 Transitions among the three states need their own transitory feedback, not only end-state cues · C4.26.4 A non-visual channel can carry at least one of the three states
• Adjacent: C4.11 Substitutes for contact feedback · C4.03 Gesture onset and offset detection
• Search: three-state model · Buxton · in-range locked activated