Shape and location distinguish controls by touch

What it is

Eyes-free use first has to tell which control the finger is on, and only then to press it. Discrimination is tactile coding: a ridge, a well, a thumbwheel, a fixed place on the left or right spoke, so the finger knows where it is at contact. Icons printed on the cap and back-lit legends do not exist when no one is looking. A row of identical square keys lets location memory collapse, and people look anyway. This entry is not about whether a key has travel. It is about what lets keys be told apart by touch.

Why it happens

The fingertip is spatially fine, but it has almost no label for “same shape in a row.” Location coding uses the body schema: left-spoke thumb rest, right-spoke thumb rest, a place behind six o’clock—three anchors that are hard to swap. Shape coding uses edges: a wheel is not a key, a bar is not a dot, a boss is not a pit. Stacked, they still correct a slightly wrong landing. A flat touch surface erases both shape and location and leaves only visual legends, so discrimination collapses when the eyes are away. Gaps that are too small fuse two edges into one and eat the resolution of location coding.

Studying it

Cover printed legends, leave only shape and place, and have people hit “volume up,” “answer,” and similar commands. Contrast identical shapes at equal spacing with mixed shapes plus left/right anchors. Blank identical caps that vary only in position are another version.

Independent variables: whether shapes differ, left/right asymmetry, gap size, presence of a raised landmark. Dependent variables: eyes-free hit rate, search time, looks-to-hand used as repair, adjacent-key errors.

Practice improves location memory, so report first-touch and after-practice separately; most drivers of a production car do not drill the same wheel every day. Gloves wipe fine pips; coarse shapes and left/right anchors survive better.

Where it stops holding

With very few keys (one on each side), shape contrast matters less and location already suffices. As the count grows, an undifferentiated array fails first. After the wheel turns, “left” in cabin coordinates is still “left” relative to the hand, but novices search in cabin space and are briefly lost. Night back-lighting that turns every key into the same bright patch helps people who look; it does nothing for people who touch. Do not treat glow as a substitute for tactile coding.

Applying it

• Every in-motion key should feel unlike its neighbor at contact: wheel versus key, boss versus well, bar versus round. Do not ship a row of same-height, same-width squares.
• Use the spokes and the natural thumb rest as anchors. Put high-frequency keys on those anchors; do not squeeze a featureless extra row between them.
• Leave a gap the fingertip can feel as “two keys.” Drop a function rather than collapsing coding resolution.
• Verify by covering all print and icons, then finding keys by spoken command in thin gloves. Every swapped pair is a shape or location that did not separate them.

Related

• Within the group: K6.04.1 Physical controls can be used without looking · K6.04.3 Control-to-function mapping must stay stable across a vehicle lineup
• Adjacent: A4.02 Two-point discrimination threshold and body-site differences · A4.05 Proprioception and kinesthesia
• Search terms: tactile coding · shape coding · location coding