NTSB's study examined accidents from 1978-90, which encompassed the dawn of crew resource management (CRM) programs. Among CRM's key tenets: respectfully questioning colleagues is part of a solid safety culture.
Before CRM, monitoring deficiencies were easily masked by an industry culture that did not promote questioning one's superiors. Investigators probing crew-related fatal accidents often did not know if a crewmember did not notice the captain's mistake, or simply did not challenge a more seasoned colleague.
Operators began embracing CRM training in the 1980s. The FAA mandated it for scheduled airlines in 1995, while non-scheduled commercial operators faced a March 2013 deadline to implement it.
Effective CRM practices make deficient monitoring easier to spot. As CRM programs took hold, it became clear that, despite numerous benefits, monitoring was not adequately addressed.
The FAA responded by updating CRM guidance (but not the rules), adding emphasis on monitoring. But in 2003, the agency, taking a cue from several carriers, replaced the term “pilot not flying” with “pilot monitoring” in guidance on developing flight-deck standard operating procedures. “It is increasingly acknowledged that it makes better sense to characterize pilots by what they are doing rather than by what they are not doing,” the agency reasoned. “[T]he term 'pilot not flying' misses the point.”
As regulators moved forward, human performance specialists examined the issue in the context of flying a modern aircraft. Among the findings: Reliable automation makes monitoring more challenging. “The human brain just isn't well designed to monitor for an event that very rarely happens,” explains Key Dismukes, the former chief scientist for aerospace human factors at NASA Ames Research Center.
Humans also struggle to detect minor changes in their environments, Dismukes says. While master caution lights present enough contrast to be noticed quickly, for instance, subtle changes on instrument panels do not grab a pilot's attention. “We're not well designed to monitor a little alphanumeric [indicator] on the panel, even though when that alphanumeric changes, it is telling us something important.”
While industry has learned much about monitoring, operators are struggling to translate that knowledge into effective procedures. In February 2007, the NTSB concluded that the flight-crew's failure to monitor airspeed was a causal factor in the 2005 crash of a Cessna Citation in Pueblo, Colo., that killed eight. The board, going beyond its 1994 recommendation, said the FAA should require monitoring in pilot-training programs. The FAA responded that CRM guidance adequately addressed the subject.
Over the next three years, inadequate monitoring would play a key role in one fatal accident and one high-profile incident. In February 2009, Colgan Air Flight 3407 went down near Buffalo, N.Y., killing 50. The NTSB determined that the crew's failure to notice a low-speed warning contributed to the stall that caused the Bombardier Q400 to crash. In February 2010, the overrun of an American Airlines 757 at Jackson Hole, Wyo., was set up in part by “a lack of pilot training emphasizing monitoring skills.”