Losing the click of contact means gestures need stronger state feedback elsewhere

What it is

A hand hangs a few centimetres off the glass. The system may already have a target, or it may have captured nothing—contact never supplied the detent of a completed press. Touchless input needs stronger state feedback means that proximity sensing, hover, depth cameras, or mid-air gestures on a public display must make “seen / aiming / committed” more conspicuous than the same states on a private touchscreen. The claim is not the channel fact that mid-air lacks haptics. It is that, in public, the signal has to be readable a step away; otherwise hygiene-driven touchless is judged “broken” on the first attempt.

Why it happens

Touch folds aiming and confirm into one press; the mechanical travel is the commit. Touchless splits the steps: enter the sensing volume, dwell on a target, then cross a confirm threshold (dwell, push, pinch). A public display adds a further uncertainty—people do not know whether they count as a user. A cursor in the corner is invisible at one to two metres, in sun, or in peripheral vision. Latency is read as “it did not see me”: the passer-by funnel loses people here because the system stayed silent, not because the glass looked dirty. Feedforward has to come first—you are already being sensed—before command-level feedback. Ambient noise and glare swallow thin visual change, so the same three states on a public screen usually need more size, more contrast, or a non-visual channel to survive the site.

Studying it

Measure “did the system see me” together with false triggers, not only final selection accuracy. Run the same task with and without an “in sensing” cue, and with short dwell versus push-to-confirm, at real walking distances.

Independent variables: sensing-state visibility, confirm method (dwell / push / gesture), feedback size and contrast, presence of sound or full-screen motion. Dependent variables: attempts before first successful confirm, share who leave believing they were unseen, passer-by false triggers, rated certainty of being sensed.

Lab participants already stand in the capture cone, so “do I count as a user” never arises; field coding must split lingering from intentional interaction. When sunlight washes out a low-contrast highlight, indoor cursor sizes do not transfer outdoors.

Where it stops holding

Handhelds and headsets put feedback on the face, so the intensity demand drops. NFC taps and card inserts carry their own mechanical or electromagnetic edge and do not need this markup. A quiet indoor guide screen at short range can survive weaker feedback. Voice has its own listening indicator; mid-air feedback strategy does not copy over. Over-eager sensing highlights also recruit passers-by as users and produce public false commits—feedback should be strong, but the sensing gate still has to separate walking past from standing still.

Applying it

• Before anyone issues a command, show with a large highlight or silhouette that they are already sensed; do not wait for the first failure to spawn a cursor.
• Give aiming and confirm different shapes so one ring of light does not mean both “acquired” and “already clicked.”
• Outdoors or in noise, add a sound or a screen-level flash on confirm; hairline motion is not enough.
• Verify by filming at the installed distance and counting reaches that get no response then leave; add sensing feedforward and compare the same traffic window, rather than quoting recognition accuracy alone.

Related

• Within the group: K7.06.1 Public touch surfaces carry hygiene concern · K7.06.3 A touch fallback path must remain available
• Adjacent: C4.11 Substitutes for Contact Feedback · C4.26 Distinguishing Three-State Feedback · K7.02 Discoverability of Interactivity
• Search terms: touchless feedback · feedforward · hover-to-select