Zoning has to be verified with eyes off the device, by touch alone

What it is

Zones that look clean on a drawing are not the same as zones a finger can use with the eyes shut. Zoning has to be verified with eyes off the device, by touch alone is an evaluation protocol, not another zoning trick: turn vision off (and often sound), start from the real grip, complete named key presses, and let wrong region, wrong key, and searching decide whether the map works. Success measured with legends visible measures reading, not zoning.

Why it happens

Once vision is present, people aim with the eyes to compensate for missing borders, missing unique silhouettes, even flush caps. Zoning’s whole claim is “the map remains when vision is absent,” so a test that allows looking never tests the claim. Sound and on-screen highlights also correct a miss after the fact, laundering a search that should have counted as failure into a success. The protocol therefore needs: vision off, non-tactile feedback delayed or off, grip matching the field, and a task of “go press this function,” not “point to where this key is.” The latter still uses spatial language; the former is eyes-free use. Gloves, vibration, and the other hand occupied are part of the protocol, not optional extras.

Studying it

Write eyes-covered (or gaze forced onto the primary task) as the default condition, not a last “accessibility add-on.”

Independent variables: vision on/off, sound on/off, gloves, primary-task load, whether the bezel may be felt first for calibration. Dependent measures: first-contact region accuracy, number of corrections, primary-task score, time.

The gap between the same layout “tested looking” and “tested not looking” is how much vision is doing zoning’s job. A small gap: zoning may be real. A large gap: the blocks on the drawing work for eyes. The sample must include people unfamiliar with this device; fluent users will use memory rather than tactile borders.

Where it stops holding

Some functions are looked-at in real use (calibration in Settings); failing their screen layout with an eyes-free protocol is too strict. A touchscreen with no physical border will fail the protocol stably—that reveals a missing channel, not a botched drawing of zones. If a safety review already requires “read the legend then press,” an eyes-free test is still useful: it tells you whether someone in a hurry will skip the legend. Children and different hand lengths change the grip; a “standard hand” in the protocol cannot be the only sample.

Applying it

• Make “eyes off, touch only” a must-pass for zoning sign-off, not a demo-day performance.
• Score first contact and wrong region; do not only score whether the right key was eventually pressed (a glance in the middle is a fail).
• Primary-task scenes (driving mock-up, viewfinding) are closer to real than a blindfold: blindfolds make people extra careful; driving is what lets old defects out.
• Verify: a task list, gaze on the primary scene, one attempt each. Wrong region or a mid-task look at the device is a fail. Redesign borders or privileged slots where fails cluster, and rerun the same protocol—do not switch to “looking allowed” to make the numbers pretty.

Related

• Same group: C10.09.1 Key zoning uses differences in shape, material, or raised edges so fingers locate regions by touch · C10.09.2 High-frequency function keys belong in the most reachable, most discriminable positions · C10.09.3 Layout changes break long-time users’ muscle memory
• Adjacent: C10.01 Tactile Localization of Physical Keys · C7.06 Hands-Free and Eyes-Free Use
• Search: eyes-free evaluation · tactile-only test · zoning validation