Voluntary blink commands are usually recognized by lid-closure duration, not a single blink

What it is

What actually pulls a voluntary blink out of spontaneous blinks in practice is lid-closure duration: closed longer than a cutoff (often 300–800 ms or more) counts as a command; an ordinary blink does not. Recognition is of a hold, not of the instant of opening and closing. It looks like dwell, but what is integrated is the lid being shut, not gaze sitting on a control; while the eyes are closed the person sees nothing, so it cannot replace dwell selection.

Why it happens

The spontaneous duration distribution is steep on the short end; the tail beyond 200 ms is thin. Putting the threshold beyond that tail drops false positives sharply; the cost is that the user must deliberately hold the lids, an isometric contraction. Vision is off during the closure; if the UI changes in those hundreds of milliseconds, it no longer matches on opening. The longer the threshold, the more it collides with protective long closures in dry eye and with slow fatigued blinks.

Some systems use “blink three times” as a command—that is count, not duration, and is harsher on Parkinson’s or on people whose blink rate is already high. Duration’s advantage is one event, one threshold; its cost is blindness while closed, so it cannot aim.

Studying it

Sweep closure thresholds, plotting false positives and misses between spontaneous viewing and prompted long closures. Also log whether the screen updated during closure and reorientation time on opening. Contrast three short blinks versus one long closure for stability on target users, including motor-impaired users. A dryness exposure is required: after a drop of artificial tears the spontaneous duration distribution moves, and the safety margin should be taken from the post-dryness distribution. Do not set the threshold only on a short session in air-conditioning.

Where it stops holding

Very dry cornea, blepharitis, and some thyroid eye disease produce spontaneous long closures, and false positives on the duration channel rise. While driving or continuously visually monitoring, a deliberate 500 ms closure is already unacceptable. People who are fully blind do not have “the picture does not match on opening,” and the duration channel is cleaner for them. Some communication systems use a long closure as the “select” on a scanning UI; that is confirm on a scan, still not gaze dwell.

Applying it

• Use a long closure, not an ordinary blink, as the command; put the threshold beyond that user’s spontaneous tail, and allow personal adjustment.
• Do not switch critical frames during closure; on opening the user should see the pre-closure state or a clearly new state, not a half-finished animation.
• Verify by remeasuring false positives after a dryness exposure, and by confirming long-blink commands are not enabled in tasks that require continuous viewing of a road or a monitor.

Related

• Same group: C8.11.1 Spontaneous physiological blinks and voluntary blinks must be distinguished · C8.11.3 Using a wink as a command requires left/right discrimination and is infeasible for some users · C8.11.4 False triggers of blink input are easily confused with fatigue and dryness
• Adjacent: C8.02 Dwell selection · C8.04 Gaze-plus-confirm combinations
• Search: long blink · blink duration · lid closure