Complacency
When things feel safe or automated, vigilance quietly erodes and warning signs get missed.
What it means
Complacency is a state of reduced vigilance and attention to a task or risk that arises when a person feels the situation is under control, safe, or reliably handled by something else, leaving them under-prepared for the moment it is not. In human-factors research it is studied especially as automation complacency, the tendency to monitor an automated system poorly and to over-trust its outputs, so that operators are slow to notice failures and may rubber-stamp automated recommendations even when they are wrong. The mechanism is partly attentional—sustained monitoring of a system that almost never fails is effortful and boring—and partly a calibration failure in which trust outruns the system's actual reliability. Complacency is dangerous precisely because it is invisible until tested: performance looks fine until the rare event that demanded the lapsed vigilance. It connects to normalization of deviance, where repeated uneventful violations make risk feel acceptable. It matters for aviation, medicine, driving, cybersecurity, and any domain where rare failures meet automated or routine safety.
The reliability paradox
The seminal demonstration comes from Parasuraman, Molloy, and Singh (1993). Operators ran a multi-task flight simulation while an automated routine watched engine gauges for them. When the automation's reliability stayed constant, detection of the failures it did miss collapsed after about twenty minutes under automation control; when its reliability varied over time, monitoring held up. The counterintuitive lesson is that steady, dependable automation induces more complacency than flaky automation, because consistency is precisely what signals that nothing new is worth watching. The lapse is not laziness but an economy of attention gone wrong: effort flows to the manual tasks that visibly demand it, and the quiet, reliable channel gets sampled less and less until the rare failure slips past.
What the evidence shows
Parasuraman and Manzey's (2010) integrative review pulled the scattered findings into one picture. Complacency reliably emerges under multiple-task load, when the monitored channel competes with hands-on work for attention; it is weak or absent when watching the automation is the operator's only job. It appears in novices and seasoned professionals alike, and neither practice nor explicit warnings dependably remove it. The same attentional dynamic feeds automation bias, the over-trust of a system's output, producing both omission errors (missing an event the automation failed to flag) and commission errors (following a wrong recommendation against contradicting evidence). Field data echo the lab: in conditionally automated driving studies, sizeable fractions of drivers are slow to reclaim control when the system silently stops working.
Is it really complacency?
The label has serious critics. Moray and Inagaki (2000) argued that most studies never measure complacency directly; they infer it from a single consequence, a missed automation failure, and then name that miss 'complacency,' which is circular. Detecting every rare signal is impossible even for a perfectly rational observer sampling optimally, so a missed failure alone proves nothing. Worse, monitoring a highly reliable system less is frequently the correct allocation of limited attention, not a defect. On this reading, 'complacency' risks becoming a blame word pinned on operators after an accident for behaviour that was reasonable given what they knew. The practical upshot is to judge monitoring against a normative sampling benchmark, not against the impossible standard of catching everything.
How it is measured
Because the state is hard to catch in the act, researchers also gauge a disposition toward it. Singh, Molloy, and Parasuraman's (1993) Complacency-Potential Rating Scale surveyed people's attitudes to everyday automated devices, extracting confidence, reliance, trust, and safety factors. Merritt and colleagues (2019) argued that the original conflated complacency with general trust, and built a newer scale centred on self-reported monitoring habits, treating complacency potential as an individual difference distinct from trust. Behavioural studies still lean on proxies: failure-detection rate, gaze dwell time on the automated display, verification effort before accepting a recommendation, each an imperfect stand-in. The measurement gap is exactly what the critics flag, since without a monitoring benchmark a low detection rate cannot by itself separate complacency from bad luck.
Examples
Drivers using advanced cruise-control systems take their eyes off the road and react late when the system unexpectedly fails to brake, because long stretches of flawless operation lulled them.
A radiographer who has watched the screening software flag every real case for two years starts glancing rather than looking, and the one tumour it misses sails through unremarked.
An accounts team stops checking supplier bank details because the payment system has never let a bad one through — until a convincing fake invoice arrives and is paid without a second look.
Pilots on a long-haul autopilot cruise let the flight-management computer fly untouched for hours; when it silently mistracks a waypoint, the crew notices minutes late because nothing had demanded a glance.
A dispatcher who trusts a routing system that is right almost every time keys in the delivery sequence it recommends even when the day's manifest flags a closed road the software never saw, letting the automated call override the evidence in plain sight.
Key references
- Merritt, S. M., Ako-Brew, A., Bryant, W. J., Staley, A., McKenna, M., Leone, A., & Shirase, L. (2019). Automation-Induced Complacency Potential: Development and Validation of a New Scale. Frontiers in Psychology, 10, 225. doi.org/10.3389/fpsyg.2019.00225
- Parasuraman, R., & Manzey, D. H. (2010). Complacency and Bias in Human Use of Automation: An Attentional Integration. Human Factors, 52(3), 381-410. doi.org/10.1177/0018720810376055
- Moray, N., & Inagaki, T. (2000). Attention and complacency. Theoretical Issues in Ergonomics Science, 1(4), 354-365. doi.org/10.1080/14639220052399159
- Parasuraman, R., Molloy, R., & Singh, I. L. (1993). Performance Consequences of Automation-Induced 'Complacency'. The International Journal of Aviation Psychology, 3(1), 1-23. doi.org/10.1207/s15327108ijap0301_1
- Singh, I. L., Molloy, R., & Parasuraman, R. (1993). Automation-Induced 'Complacency': Development of the Complacency-Potential Rating Scale. The International Journal of Aviation Psychology, 3(2), 111-122. doi.org/10.1207/s15327108ijap0302_2