Moving a control discards the muscle memory already formed
Aliases: positional constancy · motor program · spatial habit
What it is
For three months Bold sat third from the left on a toolbar. A redesign swaps it with Italic; the finger still lands on the old slot, and a document is italicised in bulk. People did see the new glyph — after the hand had already gone down. Muscle memory encodes a frequent control as a coordinate or a slot, not as a label to be read each time. Change the place and the old program still runs, now firing whatever occupies that slot.
This is memory formed and then demolished, not memory that never formed.
Why it happens
After enough repetitions, selection stops being visual search and becomes a motor program: the hand travels from its rest to a remembered relative place. The program is bound to the slot, not to the meaning. A redesign that only swaps the contents of slots is executed faithfully, treating the new content as the old command. That is a slip in Reason’s sense: the intention is still Bold, the movement lands on Italic.
The less people look, the more they are in a hurry, the more they are talking at the same time, the more the program outruns reading. Moving a control therefore hurts experts more than novices — novices are still reading labels; the expert’s hand has already left.
Studying it
Train a stable layout until the action is automatic (reaction time down, gaze no longer hitting the label), then swap the target with a neighbour and measure the next trials.
Independent variables: number of practice trials before the swap, adjacent versus distant swap, presence of a motion or marker cue at the change. Dependent variables: landings on the old slot, time to the first successful hit of the new place, whether gaze returns to the label.
The automation has to be in place before the swap. Too little practice and the measurement is just relearning, not discarded memory.
Where it stops holding
Rare entries have no program to discard; moving them is almost silent. A layout the user arranged themselves is a program they wrote; a product update should not overwrite it. Cross-device sync that pastes a phone’s slots onto a desktop will contaminate two programs. A visual restyle that leaves coordinates alone (new glyph, new colour) usually still hits; this claim is not about that.
Applying it
- Treat a published frequent slot as a public API: add new entries into empty space; do not insert in front of an existing frequent item and shift every later slot.
- When a move is unavoidable, leave a “moved” ghost or a jump in the old slot for a version or two, rather than swapping on the night.
- Toolbar reshuffles, tab-bar icon order, settings pages that migrate a frequent item into another group — review each as “does this kill a slot”.
- How to check: people who can already hit the control without looking, immediately after the change, no time to read. A hand that still lands on the old slot is discarded memory on the scene.