Mission impossible? Study reveals challenges of watching computers fly the plane

October 1, 2015 by Shelly Leachman
Mission impossible? Study reveals challenges of watching computers fly the plane
The challenges faced by airplane pilots charged with monitoring automated functions in the cockpit are the focus of a new study by UCSB’s Jonathan Schooler, a professor of pscyhological and brain sciences, and NASA research psychologist Steve Casner. Credit: iStock/Carlosphotos

When a commercial jet crashed on landing in San Francisco, it was ultimately determined that the plane had slowed to an unsafe speed during approach—and no one in the cockpit noticed in time to prevent the accident.

With pilots assigned to monitoring aircraft position, speed, altitude and an array of other automated functions as they appear on a computer screen during flight, how could such a major lapse occur? Easily, suggests a new study, which posits that the nature of the task itself is a recipe for failure.

The NASA-initiated scientific examination paired its own research psychologist, Steve Casner, with Jonathan Schooler, a professor of psychological and brain sciences at UC Santa Barbara. Their results indicate that humans may be inherently bad at watching computers work—and that they're unlikely to get any better, no matter how careful the selection or training.

The paper, "Vigilance impossible: Diligence, distraction, and daydreaming all lead to failures in practical monitoring task," is published in the current issue of the journal Consciousness and Cognition.

"Our study really does suggest that vigilance is a very difficult task for people," explained Schooler. "Extended uninterrupted monitoring can be draining. The antidote to that is interruptions that break up the monotony, but we also found that the interruptions themselves contributed to lapses. And people will spontaneously mind-wander, and that can also contribute to monitoring difficulties.

"So staring is draining, plus things come up which interfere with our ability to monitor, and our mind leaves the premises even when none of the other things are an issue," Schooler added. "It's a trifecta of things working against effective monitoring."

Schooler, as mind scientist, and aviation psychologist and pilot Casner teamed up to examine why monitoring failures happen even among experienced and highly trained airline pilots. For their study, they asked 16 commercial jet pilots to monitor the progress of a simulated routine flight in which high levels of cockpit automation handled the tasks of navigating and steering the airplane.

Knowing that past experiments had shown monitoring—essentially sitting and staring at computer screens—to be a tiring process that quickly leads to fatigue and inattention, Casner and Schooler wanted to know how the pros do it: Did they have strategies to fend off the fatiguing effects of long watches?

The researchers found that the cockpit environment is busy enough that pilots were often sidetracked by other tasks, such as talking to or configuring the airplane's systems, which curtailed fatigue. But they also discovered that these pop-up tasks could themselves cause pilots to miss important events during flight.

Most interesting, Schooler and Casner said, is what happened when the pilots weren't interrupted. Rather than focusing solely on monitoring the flight, they instead created their own distractions by engaging in what the researchers call "mind wandering."

When periodically asked, during the study, what they were thinking about, pilots admitted to thinking "task-unrelated thoughts" up to 50 percent of the time—mental excursions that frequently led to missed events in flight. All in all, the researchers found, pilots missed 25 percent of all altitude crossings they were charged with monitoring.

The pilots tried to limit their mind wandering to times when there were fewer demands on their attention, Schooler said, but ultimately found that even in circumstances with high demand they routinely mind wandered.

"We had anticipated that the longer people monitored that in itself would be associated with inferior performance, and we actually didn't observe that particular pattern," Schooler explained. "But we had failed to appreciate how the solution—the interruption—itself compounds the problem. We were also surprised at the sheer number of times that missed the altitude callouts and the high frequency with which they admitted to mind wandering.

"We should be very wary of relying on people to serve in a monitoring capacity—especially now, when we do have technology that can fill the monitoring role," he concluded.

Casner concurred, suggesting that real-world monitors may be caught between a continuous-vigilance approach that is doomed to fail, a dynamic environment that cannot be fully controlled and what may be an irresistible urge to let one's thoughts drift.

"This task of watching over a computer system while it works is incredibly trying, if not impossible, for a human being to do well," Casner said. "You can try paying attention, and you can try taking brief breaks, but sooner or later you'll miss something important. This is a job for a robot, not a human being. It's time to rethink the way we design these systems. Let the people do the stuff they are good at, and let the computers handle the mundane chores."

Explore further: Staying on task in the automated cockpit

Related Stories

Staying on task in the automated cockpit

May 6, 2014

Automation in the cockpit is traditionally believed to free pilots' attention from mundane flight tasks and allow them to focus on the big picture or prepare for any unexpected events during flight. However, a new study published ...

How prepared is your pilot to deal with an emergency?

January 14, 2015

Pilots receive extensive training on effectively handling emergency situations that may arise in the cockpit. However, during several recent highly-publicized airline incidents, the pilots neglected to apply the skills they ...

Image: View from an F-15D

May 28, 2015

NASA pilot Jim Less and photographer Jim Ross pull their F-15D #897 aircraft away from a KC-135 refueling tanker. NASA is supporting the Edwards Air Force Base F-15 program with safety and photo chase expertise.

Recommended for you

Dutch open 'world's first 3D-printed bridge'

October 17, 2017

Dutch officials toasted on Tuesday the opening of what is being called the world's first 3D-printed concrete bridge, which is primarily meant to be used by cyclists.


Adjust slider to filter visible comments by rank

Display comments: newest first

not rated yet Oct 01, 2015
If I remember one incident correctly, the plane did exactly what it was programmed to but one altimeter was wrong causing it to believe it had 'landed' in mid air and drop like a brick the last few hundred feet, with the auto-throttled engines then too slow to rev up in time. It was Programmed to avoid one situation on the ground and caused another in the air: While performing what looked exactly like a routine textbook approach. Only the airspeed would have given it away. But the stall warning came too late to rev the engines back to power.

The pilot has way too may things to do if an approach gets complicated, to cross check instruments and second guess the avionics when near to the ground.
Modern avionics should cross check the three altimeters (and all the other duplicated/backup instruments) and get the pilot involved sooner.
5 / 5 (1) Oct 01, 2015
Id agree that watching a computer do the same thing over extended period of time can put you to sleep, but I disagree that that is any excuse for failure during landing procedure. It is just not a long enough period of time to lose attention.
5 / 5 (3) Oct 01, 2015
"It was ultimately determined that the plane had slowed to an unsafe speed during approach—and no one in the cockpit noticed in time to prevent the accident."

This is a VERY over-simplified explanation of the cause of the crash. There were several pilot-initiated actions that preceded it "slowing to an unsafe speed." I'm not trying to say the findings from this study are wrong or useless... they are good data points.

I guess I just don't like that this write-up almost implies everything was "computer controlled" up until the crash and the pilots just didn't notice that things weren't right. Yes, there were things they didn't notice... but there were actions they did that led to the situation in the first place.

5 / 5 (3) Oct 01, 2015
Before you can watch a computer fly an aircraft, you need basic stick and rudder skills.

However, there are almost no private aviation business in Asian countries. The pilots graduate straight from the military to the airlines. They get virtually no chance to learn plain stick and rudder flying. Furthermore, they also have a very strong taboo against criticizing their seniors. One pilot in that Asiana airliner did in fact realize what was going on, but he was too timid to speak up forcefully because he was one of the younger members.

It wasn't just the automation that failed there. It was culture, training, inexperience, and fatigue. That said, the Airbus versus Boeing philosophies are interesting contrasts that every automation engineer should study.
1 / 5 (1) Oct 02, 2015
Maybe a big red flashing light and a recorded voice shouting "Wake up! It's crash time.".
not rated yet Oct 03, 2015
Besides "ALL THE TIME" there are very important periods for the pilots to be paying attention:
- push off from the gate and taxi out to the run way and take off
- and approach, perform landing and taxi to the gate

If a pilot can't muster the wherewithal to accomplish these things, I wouldn't trust them to pay attention anywhere.

Yet despite this, flying is safer than driving. Plane crashes gather more attention than car crashes though, since they can potentially affect hundreds of lives, not just a handful.
not rated yet Oct 03, 2015
Haptics could be useful here.
5 / 5 (1) Oct 04, 2015
Haptics could be useful here.

Haptics already ARE in use. They were pioneered by the aviation industry decades ago. Where do you think the aerodynamic stall stick shaker came from?

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.