Pilots make mistakes. Mistakes are a product of the brain. If it were possible to identify the common neurological precursors for pilot errors, it might be possible to prevent them.
That, at least, is the theory on which the Institut Supérieur de l'Aéronautique et de l'Espace in Toulouse is working. Prof Frédéric Dehais, who holds the AXA chair of neuroergonomics for flight safety at ISAE Supaero, describes the research programme he is heading as a merger of neuroscience, human factors and computer science.
A Boeing study quoted by Dehais shows that, between 2002 and 2011, 1,493 people died in some 50 in-flight loss-of-control or controlled flight into terrain accidents. “In the majority of these stressful but recoverable situations”, he says, “it appeared that the crews were unable to identify the problem, and continued to take irrational decisions that did not make sense according to what was happening.” Even the intervention of audible alerts made no difference to the crew’s persistence with a doomed strategy.
“Accident analysis reveals that the complexity of modern transport aircraft can overwhelm the most experienced crews when something goes wrong,” Dehais says. “They can become confused, stressed, and fail to assess the criticality of the situation. They also persist in erroneous courses of action despite auditory and visual warnings in the cockpit. In the end, they crash.”
He adds: “These dramatic events are surprising because the pilots are supposed to be highly trained. There is a need to understand what kind of neural mechanisms are leading to these situations.”
This programme uses a unique range of neural measurement techniques while putting pilots through flying tasks in a flight simulator. “The purpose of my research is to uncover the underlying neural mechanisms of human error that lead pilots to persist in irrational behaviours,” Dehais says. Serious accidents may be rare, but 70% of them result from human error, so a deeper understanding of the nature of error and what causes it is an important objective.
“Our approach combines cutting-edge brain imaging, signal processing and artificial intelligence techniques as well as a unique methodology from basic protocols to ecological experiments,” explains Dehais. “Results from this research will be strongly beneficial to aviation safety, leading to the implementation of innovative solutions to mitigate human error.”
Tools used to monitor pilot reactions during flight simulator exercises – and real flights – include eye-tracking, measurement of pupil dilation, electro-encephalograms to track deep brain activity, and infrared sensors that can show activity in critical surface brain areas, as well as advanced brain imaging and signal processing techniques.
Pupil dilation is a reliable external indicator of mental stress, says Dehais, and there are now very precise ways to monitor it. The scientists can watch as a highly stressed pilot’s brain literally shuts down many of its critical faculties and shifts from rational decision to emotional reaction. At that point the pilot is in a state of “inattentional deafness”, where audible alarms and spoken instructions are ignored.
More conventional physiological measurements are also used. The heart rate indicates the levels of mental stress and physical workload. For example, a take-off in a simulator does not raise the heart rate to the degree it does in a real aircraft.
Wired-up pilots flying a real light aircraft are not forced to undergo an actual engine failure, but the throttle is pulled back and the pilot has to look for a forced landing opportunity. Meanwhile his or her brain activity is being measured and recorded so common neurological activity patterns can be identified and related to specific activities and stress levels.
All the same monitoring techniques can also be used for the pilots of remotely operated unmanned aerial systems.
Backed by the AXA Research Fund, the ISAE is working to identify the patterns of neurological activity that occur when pilots become confused, overloaded, or focused on non-critical inputs to the exclusion of critical ones. If one can identify and understand the neurological reaction, says Dehais, it is possible to apply what he calls “cognitive countermeasures”.
In the AXA-ISAE project, Dehais is working in parallel with research being carried out at NASA’s Ames Research Center, Stanford and Harvard universities, the Centre for Information and Neural Networks in Japan, and Laval University in Canada.
Artificial intelligence may still be thought of as being in the realms of science fiction, but ISAE believes it may be used in the recognition of overload and the timely deployment of appropriate cognitive countermeasures.
Even factors such as “emotional bias” can be recognised. ISAE says this is caused by pressures from “outside the cockpit”, such as pilot perception of commercial pressure to land when a go-around would be wiser. The ISAE has tested the cognitive effect of emotional bias by offering pilots a financial benefit to land versus a small financial penalty for a go-around, and have observed in their brain a greater degree of neurological stress during the decision-making process, because it is no longer a decision affected purely by operational considerations. The effects on brain activity of ambient light, fatigue and age can also be identified, as can the effectiveness of mitigating techniques for each of those.
Knowing, as neurologists now do, which parts of the brain control specific functions, the ISAE scientists can predict which tasks the pilot will no longer be able to perform as the stress increases.
There are traditional ways of helping pilots deal with stress. Good training and high levels of knowledge in a pilot mean that he or she is less stressed by any given situation than a pilot with poor training and knowledge, and the ISAE says this is visibly true. But the institute is looking beyond the traditional to what else will be effective.
This new level of neuroergonomic understanding promises to enable manufacturers to eliminate alerts and other stimuli that do not work, and develop completely new “cognitive countermeasures” designed to attract the attention of pilots whose cognitive capacity has been swamped.
An example of such a system is a window appearing on the navigation display showing an animation of a pilot carrying out the required action to ensure recovery. Neurologists know that, when all else fails, they can invoke the human imitative function – the one that makes a person yawn when someone else yawns.
Another “cognitive countermeasure” that the ISAE has found to be highly effective is, when a pilot is clearly focusing on one information source to the exclusion of other critical ones, to remove that information completely. The pilot’s instrument panel scan immediately switches elsewhere looking for information to replace the removed data, and discovers important information he or she has been missing.
Previous approaches to dealing with common mistakes have included multiplying the number of alerts, including audible or spoken alerts. ISAE explains that, beyond a certain point, this can be at best pointless and at worst counterproductive, increasing stress and narrowing a pilot’s focus further. So it makes sense to understand what happens to pilots’ cognitive functions when they are stressed.
Neuroergonomics has enormous potential. It could be used during the pilot selection process, and also for monitoring the effectiveness of training techniques or of different cockpit designs.
The ability to identify “emotional bias” could eliminate speculation about whether pilots who work for airlines that require them to be self-employed and paid by the hour are under more stress than pilots with a salary and incentives. The accuracy of the observations is such that it might be able to indicate the kind of mistakes pilots under emotional stress are more likely to make.
Dehais points out that the ISAE research has applications to other people who work in high stress, safety-critical jobs, such as surgeons and those at the front line in nuclear power control.
Whatever else the ISAE’s continuing research discovers, this project has huge potential to advance mankind’s understanding of aviation human factors.