Vigilance decrement

Vigilance decrement is the decline in detection performance that occurs during prolonged periods of monitoring for rare or infrequent signals. First documented by Norman Mackworth in 1950 using the 'clock test' — a task in which observers watched a clock hand and pressed a button when it made a rare double jump — the phenomenon has been replicated across a wide range of domains, including radar monitoring, medical alarm surveillance, and nuclear power plant control. The decrement is steep: detection rates can drop by 50% or more within the first 30 minutes of monitoring.

The standard explanation is that sustained attention is a resource-limited capacity that fatigues over time. More recent research suggests that the problem is not attentional capacity but the absence of a meaningful task structure. When monitoring is passive and the signal-to-noise ratio is low, the observer's mental model of the system degrades, and the cognitive system disengages. This is why vigilance decrement is worse in high-automation environments: the automation removes the task structure that would otherwise keep the operator engaged.

Vigilance decrement is a component of the broader phenomenon of out-of-the-loop unfamiliarity, but it is distinct. OOTLU is a loss of diagnostic competence; vigilance decrement is a loss of detection capacity. The two interact catastrophically in automated systems, where the operator is both unlikely to detect a problem and unlikely to understand it once detected. The design implication is not to ask humans to monitor better, but to design systems that do not require humans to perform monitoring tasks for which they are structurally unsuited.