Long uneventful periods weaken supervision
Aliases: vigilance decrement · rare-event monitoring · supervisory disengagement
What it is
Vigilance decrement in supervision is slower or missed detection of rare anomalies during prolonged monitoring of low-event autonomy. It is a predictable human limit of sparse, monotonous, low-action attention, not personal laziness.
Why it happens
Low event rate lowers the anomaly prior and makes subtle change resemble noise. Passive monitoring reduces active task-model maintenance, while reliable automation teaches that looking is unnecessary. An event then requires detection, context recovery, authority understanding, and response.
Studying it
Long-duration studies should vary event rate, salience, active checks, and secondary tasks, measuring misses, response-time slope, awareness, gaze, and recovery. Short or event-rich trials remove the phenomenon. Frequent false alarms create alarm fatigue instead.
Where it stops holding
Training and motivation help but do not eliminate decrement. Event-driven notification shifts dependence to the detector. “Pay attention” is not a high-risk control, and shift rotation only works with sound handoff.
Applying it
- Use event-driven supervision plus periodic active checks that resample critical state.
- Design rotation, breaks, handoff, and limits on systems per person.
- Test at real shift duration with rare anomalies, analysing miss and latency curves alongside false alarms and notification failure.