Slip
The intention is correct, but the hand activates the neighboring control during motion, vibration, or high workload.
Design responseIncrease spacing, tactile contrast, guards, and easy reversal.
An interactive study in attention under pressure
The200-mphInterface
A Formula 1 steering wheel is a decision surface operated by touch, under load, with little attention to spare. Explore how it reduces demand, then feel what divided attention does to your own performance.
02Hands available
20+Controls within reach
FINITEAttention to divide
01 / EXPLORE
Control without looking
Tap any labeled control. Its design logic appears beside the wheel.
Tap a control
Eight controls, one rule: every input must work without a glance.
02 / EXPERIENCE
Your attention has a limit
Match eight letters. Then repeat while tracking a moving target. The change in your response time is the dual-task cost of divided attention — what it does to performance. How demanding it feels is a separate question: that's workload, measured with tools like NASA-TLX.
Session Ready
0 / 16 prompts
Two rounds · eight prompts each
See it. Match it.
AASee a letter. Press the same letter.
Round two adds one more job: keep the cyan dot inside the yellow zone.
Continuous tracking taskDrag or use arrow keys
See a letter → press the same letter
GET READY
Wait for the cue.
Your cockpit debrief
The dual-task cost
Baseline median—response time
Dual-task median—response time
Dual-task tracking—time on target
You did this sitting still, in silence.
Divided attention is not a character flaw. It is a design constraint. Good interfaces reduce the number of choices, keep controls predictable, and make it easy to catch and undo mistakes before they become consequences.
03 / UNDERSTAND
“User error” is not a diagnosis
Four common mechanisms can produce the same wrong action. Each one needs a different fix.
Lapse
The intention is lost before the action — a skipped step in a sequence, a setting left unchanged, a mind gone blank under load.
Design responseChecklists, forcing functions, and reminders at the point of action.
Mistake
The action matches the driver's plan, but the plan is wrong because the mental model or available information is incomplete.
Design responseClarify system logic, improve training, and surface consequences before commitment.
Mode error
The control does what it was designed to do—but in a different active mode than the driver believes.
Design responseMake state conspicuous, reduce hidden modes, and require confirmation for costly transitions.
04 / TRANSFER
Same science, different stakes
F1 makes the constraints visible. Diabetes technology makes their consequences personal: state, guardrails, alerts, and recovery.
Visible wheel mode
→
Clear insulin-delivery state
A correct input in the wrong context can produce the wrong outcome. The active workflow, entered value, unit, and pending effect should remain visible through review and confirmation.
Pit-limiter constraint
→
Guardrails and automation
A safety constraint can interrupt a hazardous path. Its value depends on clear feedback: the person needs to know what the system did, why it acted, and what still requires attention. The trade-off is real — every guardrail costs speed and flexibility — and worth paying when the error costs more than the delay.
Short radio messages
→
Alerts with one clear next action
Urgent and routine notifications must be easy to distinguish, including when someone is tired or distracted. Priority, plain language, and acknowledgment protect attention and trust.
Practiced recovery
→
Training for interruptions
Training should cover more than the ideal workflow. People need realistic practice recovering from an alert, an interrupted sequence, or an unexpected reading before facing it alone. But training is the weakest safeguard: design the error out first, and train for what design can't cover.
The method is the bridge: identify critical tasks, test them under realistic distraction and fatigue, observe slips and mode errors, trace each failure to its conditions, then redesign the interface—not the person. In medical devices this discipline is formalized as usability engineering: IEC 62366-1 and FDA human-factors guidance require identifying use-related risks, testing designs with users early and often, and validating that representative users can perform critical tasks safely — down to how alarms compete for attention (IEC 60601-1-8).
The sharpest test: design for the person having a hypoglycemic episode — shaky, confused, cognitively impaired — not the ideal user. If the interface works for them at their worst, it works for everyone at their best.
FinishTake the principle with you↓Speed makes the constraint visible.
Human factors makes the response humane.
The best interface is not the one that asks a perfect person to perform perfectly. It is the one that still supports the real person—moving fast, carrying context, sometimes getting it wrong, and still deserving to be safe.