Automation in the cockpit is making pilots’ thinking skills duller.

Automation Is Making Pilots’ Thinking Skills Deteriorate. But Laziness Isn’t to Blame.

Automation Is Making Pilots’ Thinking Skills Deteriorate. But Laziness Isn’t to Blame.

The citizen’s guide to the future.
Dec. 12 2014 7:47 AM
FROM SLATE, NEW AMERICA, AND ASU

Dumbing It Down in the Cockpit

As automation gets sharper, pilots’ thinking skills are getting duller.

Photo by Harry & Ewing courtesy of The Library of Congress
Airplanes of yesteryear presented more of a challenge to pilots than today’s planes, but that may not have been all bad.

Photo by Harry & Ewing courtesy of The Library of Congress

Long gone are the leather jackets, goggles, and silk scarves flung over the shoulders of aviators who wrestled with flight controls, furiously scanned instruments, and navigated using paper charts. Airplanes have been largely flying themselves since the early 1980s. Today, with a few keystrokes, pilots program the details of a flight route into a computer that calculates, in seconds, the bearings, distances, altitudes, speeds, and fuel needed to get the airplane from city to city. Aircraft navigation? Yes, there is an app for that.

Once airborne, pilots can use an autopilot that automatically calculates and carries out the control inputs needed to guide the airplane along the programmed route. Other systems warn pilots about airplanes that come too close, or a nearby mountaintop or thunderstorm. When aircraft components fail, sensors throughout the aircraft detect the problem and tell pilots which steps are needed to remedy it. And when the tires hit the runway at the destination airport, the autobrakes smoothly bring the airplane to a stop.

All this automation in the cockpit has prompted researchers like me to wonder what might be happening to pilots’ ability to fly the airplane “the old-fashioned way” when the automated systems go awry.

Advertisement

In 2010 Matt Ebbatson, a researcher at Cranfield University, was first to investigate this when he asked a group of pilots to manually steer a Boeing 737 along a precision approach to a runway using no automation whatsoever—a rare request in today’s automation age. Ebbatson found pilots’ stick-and-rudder skills to be noticeably rusty but not to the extent where any crashes ensued. Pilots drifted a little to the left, a little to the right, but ultimately made it to the airport in one piece.

I read Ebbatson’s paper with great interest. I wondered whether he had just shown the world that flying an airplane is “just like riding a bike”—maybe we didn’t need to worry too much about pilots’ skills being dulled by automation. But my airline pilot collaborator Richard Geven and I doubted this conclusion after we more closely examined recent airline accidents that were called out as examples of deteriorating pilot skills. In the case of Continental Connection (Colgan) Flight 3407 in early 2009, the flight crew encountered an aerodynamic stall, an abnormal situation that both pilots had recently practiced during their annual training. But when the stall happened in real life, the pilots did the opposite of what they were trained to do and the ensuing crash claimed 50 lives. These actions left investigators with the question of how the flight crew could have misinterpreted such a familiar situation.

In July of 2013, the crew of Asiana Flight 214 allowed their attention to lapse during an approach to San Francisco International Airport. As the pilot manually steered the airplane, he relied on an automated system to control the airplane’s speed. For reasons the crew misunderstood, that system allowed the airplane’s speed to drop dangerously low. By the time the crew noticed the problem, it was too late and the airplane crashed short of the runway.

After Geven and I looked at these reports of misinterpreted situations, lapses in pilot attention, and confusion about the automation, we wondered whether the problems we were seeing with manual flying had less to do with what pilots did with their hands and more to do with what was going through their heads. So we decided to do our own study at NASA to find out.

Advertisement

We invited 16 Boeing 747-400 pilots to come to NASA Ames Research Center and fly our full-motion simulator. We asked them to fly in ways that would separately test both their stick-and-rudder skills and the cognitive skills that pilots use to think their way through situations such as the one encountered during recent accidents. Like Ebbatson, we found pilots’ stick-and-rudder skills to be a bit out of shape but mostly intact. However, when our pilots were asked to interpret something out of the ordinary, mentally keep track of where they were, and understand how and why the automation works, their performance quickly slumped. Four of our 16 pilots failed to keep track of their position and missed the airport altogether. Eight of 16 pilots placed the airplane in a near-stall condition four times or more before figuring out that their airspeed indicator had become unreliable.

When we were designing our study, our colleague Jonathan Schooler, an expert on mindfulness, proposed that some pilots’ thinking skills might have deteriorated more than others. So at his suggestion, we periodically asked pilots what they were thinking about as they flew. Pilots indicated whether they were thinking about something related to the flight or about something unrelated to the flight, such as dinner or a vacation. And out of these data jumped the biggest result of our study. We found that pilots who habitually let the computers do the thinking and who allow their thoughts to drift during flight were more likely to have deteriorated cognitive skills.

It’s tempting to call this laziness, but that’s not exactly right. Mind-wandering experts like Jonathan Smallwood are quick to point out that our minds are restless. When we’re not given something stimulating to think about, our minds naturally drift onto something else that is. If someone asks you to monitor a light that is known to turn red once every few hundred hours, you find it mostly impossible to not think about something else more pressing or interesting. Many have pointed out that “sitting and staring” at a computer that does our job for us is not something that creative, interactive, problem-solving humans are cut out to do. As Boeing 757 pilot Helena Reidemar bluntly states: “We’re not robots.”

So how do we keep pilots’ thinking skills fresh? One idea we are currently exploring is an app that challenges pilots with unusual and thought-provoking flight scenarios. These scenarios could explore situations that we hope would never happen in real life but for which we would like pilots to be ready nonetheless. And unlike costly formal training done in high-fidelity simulators, pilots could sharpen their thinking skills anytime, anywhere using a laptop or mobile device. One glaring limitation of our study is that we tested only a handful of needed cognitive skills (like finding the airport). The design of a learning app would require a more thorough study of which thinking skills have fallen to disuse as a result of automation in the cockpit. And then comes the question of how to introduce it to pilots. Enacting new training regulations is a long, complicated process. Maybe game designers could help us design something that pilots find irresistibly fun: How about Angry Pilots?

But even if we manage to avoid the deterioration of cognitive skills, pilots are still being dumbed down in another way. With each new revision, the computers in the cockpit are becoming more complex. Edwin Hutchins, a cognitive scientist who teaches (and surfs) at the University of California–San Diego, warns of a Catch-22 situation in which automation is imperfect enough to require pilots to take over when problems arise, but too complex for pilots to understand and too reliable for them to successfully stare at for any length of time. If Hutchins is right, we seem to be left with two possibilities. We could design automation that flies the airplane with pilots instead of for them: systems that work cooperatively with humans who remain meaningfully engaged in the process. The other possibility is to make the automation more sophisticated and eliminate the need for human pilots. But such a march toward total automation will take time. Meanwhile, as we watch the “flying IQ” of the automation surpass our own, pilots may have little choice but to press a few buttons, hope for the best, and think about something else for a while.

This article is part of Future Tense, a collaboration among Arizona State University, New America, and Slate. Future Tense explores the ways emerging technologies affect society, policy, and culture. To read more, visit the Future Tense blog and the Future Tense home page. You can also follow us on Twitter.