A non-visual channel can carry at least one of the three states

What it is

The three states do not all have to be drawn on the screen. Handing at least one of them to audition or haptics—a short click only on activated, while vision keeps a shape difference between in-range and locked—is a channel split that protects discriminability. Mid-air lacks contact, so non-visual channels are often used to fake a “press.” The claim here is narrower: they are used to pry the three states apart so all three are not stuffed into one highlight. It is not a demand that every primary state have a non-visual backup; it is that at least one state leave the visual layer.

Why it happens

Vision is already carrying content, a cursor, and possibly a boundary cue; squeezing three look-differences in tends to collapse back onto an intensity axis. Auditory and haptic pulses are sharp in time, which fits a one-shot settlement such as activated, or the grab instant of entering lock. A steady “in-range” as a continuous low tone becomes noise; vision is fitter there. After that allocation, even if a visual difference is eaten by the background, settlement can still be heard or felt. Once a channel is bound to a state (a click is always activate, never lock), users can follow progress without watching chrome. If the binding wanders by scene, the non-visual channel becomes a new puzzle.

Studying it

Compare all-visual three-state, activate as a short tone, lock as a short haptic, and tone plus haptic. Include a condition that forces looking at content (reading digits, aiming at a distant label). Dependent measures are three-state judgment accuracy, missed activations, and annoyance ratings for sound/haptics. Test eyes-closed or with contrast crushed until the UI is almost invisible, and see whether the state on the non-visual channel still survives. Treat ambient noise and gloves as failure conditions for audition and haptics; do not declare success only in a quiet lab.

Where it stops holding

In a quiet public place (museum, carriage) the social cost of sound is high; a click may be unusable and should move to haptics or into headphones. A large kiosk with no actuator has no haptics. Hearing or tactile impairment drops users off the state that was assigned there; that state must be foldable back to vision. Encoding steady in-range as a continuous tone becomes noise quickly; non-visual is fitter for events than for long steady states.

Applying it

• Pick one state (usually activate, or enter-lock) and bind it to one short pulse, off-able in settings, with a visual equivalent.
• Do not use the same click for activate and for error; the non-visual channel has to keep the three states separable too.
• Accept with “submit while looking at content”: the activate instant should be heard or felt. Add a muted / no-haptics pass and confirm the visual fallback still splits that state.

Related

• Same group: C4.26.1 In-range, locked, and activated need three distinguishable feedback forms · 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
• Adjacent: C4.11 Substitutes for contact feedback · C4.29 Visibility of capture state
• Search: nonvisual · haptic click · auditory earcon