The question
Most mental-fatigue research happens in labs with contrived tasks. But people actually get tired at real desks doing real work. Can you detect mental fatigue in the wild — unobtrusively, without strapping sensors to anyone — just by watching how they type?
What they did
Researchers at the University of Groningen logged keystroke dynamics from university employees during their normal office work for 6 consecutive weeks. They analysed typing performance across three timescales: time-on-task (how long you’ve been working), time-of-day, and day-of-week. Crucially, they tracked not just typing speed, but also error correction behaviour — how often people backspaced and retyped.
What they found
- Morning strategy: workers protected speed at the cost of accuracy. Typing stayed fast, but more errors crept in and had to be corrected.
- Afternoon shift: the strategy flipped. Both speed and accuracy declined as mental fatigue accumulated.
- Day-of-week effects: on Mondays and Fridays, people favoured speed; midweek, they favoured accuracy.
- Recovery: fatigue built up within a day, but there was no cross-day buildup — workers successfully recovered between workdays.
The key idea
Mental fatigue is visible in how you type long before it's visible in how much you type. The speed-versus-accuracy trade-off shifts in predictable ways as attention degrades — detectable in ordinary keyboard activity, with no extra hardware.
How this shapes Pomodoro tAImer
The app watches your keystroke cadence and error-correction rhythm during a work session. When the signature starts to look fatigued, it nudges the session toward a break rather than pushing you to grind through the last five minutes. This paper was one of the direct inspirations for that feature.