Pilot error – What can neuroscience teach us ?

As a student pilot in Australia I am constantly out to improve my awareness of what can kill me. It sounds harsh but that is the reason why aviation is the safest mode of transport in the world, pilots are perfectionists, we are conditioned from our Trial Introductory Flight to follow the book, do it right first time and every time. Unfortunately, pilots, no matter what they think, are human beings and make mistakes. An important aspect of human factors is dealing with these failures. An article posted on Wired.com by Jonah Lehrer, ‘Accept Defeat: The Neuroscience of Screwing Up’, caught my interest. Mistakes are a primeval thing; surely by understanding the brain we can lessen the impact or even prevent them from happening.

Lehrer uses a scientific study conducted by Kevin Dunbar into what essentially is the science of science. Dunbar’s research discovered that less than 50% of hypotheses were proven correct. In actual fact ‘experiments rarely tell us what we expect. That’s the dirty secret of science’. He found though, that instead of seeing failure, the scientists would explain away the data that didn’t fit as malfunctions. Despite this, science still somehow manages to get results. Dunbar realised that it was not in the lab that scientists had their Eureka moment but in round table discussion, after lengthy debate and brainstorming. He saw that not all data was equal, once a theory was known, data was skewed to fit that theory. Belief in a hypothesis led to blindness.
Whenever a pilot encounters a problem they create a hypothesis of what is wrong, by using data such as gauges, sounds, sights and smells and reconciling them with past experiences and training, a pilot decides on a hypothesis and begins to work to remedy it. If what Dunbar is saying is true, that 50% of hypotheses are wrong how often do pilots misdiagnose failure?

An incident that comes to mind where this possibly occurred was Air Transat Flight 236. The aircraft had developed a fuel leak in the right engine. The pilots noticed this as a fuel imbalance and opened a cross-feed valve, pumping fuel from the left tank to the right. This fuel went straight out the leak. Their hypothesis was wrong; they had misdiagnosed the problem and compounded it, leading to a loss of both engines due to fuel starvation. The pilots managed to glide the aircraft to an airport so there is no questioning their skill. It is easy for me to sit here and criticise but the question must be asked, could this have been avoided?

Lehrer proposes a four step process for scientists to learn from failure.

1. Check your assumptions: Is the hypothesis wrong not the experiment?
2. Seek out the ignorant: explain your hypothesis in simple terms to see it in a new light
3. Encourage diversity: If everyone is the same, everyone has the same assumptions
4. Beware of failure blindness: It’s normal to filter out information that contradicts our preconceptions

This is to me is essentially what crew resource management is all about. Unlike scientists though, pilots do not have the luxury of time to conduct a round table discussion. Not only that a large portion of pilots do not even have the luxury of a ‘crew’. I then have my own four step process for single pilots to learn from and act to prevent failure.

1. Train the brain to be aware: A good pilot should be aware of everything happening in and around his plane – this is easier said than done, only practice can make this skill a reality. Being able to keep track of fuel usage or engine temperature across a flight takes effort, especially when you have to remember to fly the plane as well!
2. Be able to quickly diagnose a failure and act upon it in flight: When you see things going wrong you need to be able to know what that means. A hypothesis must be created. Training and talking with more experienced pilots on the ground can aid in an accurate hypothesis being created in flight.
3. Be open to a change in hypothesis: Your initial assessment may not be right, it is possible the Air Transat Flight may have been able to conserve fuel if the pilots had noticed that fuel was depleting from one tank unusually fast. As often as time allows reassess your hypothesis, similar to you constantly checking you can make the field in a force landing, more data could become available that changes your mind.
4. Conduct a thorough analysis once on the ground: Now is the time for in depth thought; talk to other pilots, rethink your actions, did you miss something obvious? Could you have picked it up earlier?

Overall I think the most important part of this article is to recognise that knowledge (i.e.; what you are taught) can lead to blindness. The Air Transat pilots were taught to open the cross-feed valve when there is a fuel imbalance and didn’t consider another hypothesis. To me this is a lesson I will store away, have in my sub-conscious and be aware of in case one day the worst happens.

This is a guest post from Rob Close, I hope you liked it as I did. Please leave your comments as usual, and if you feel like becoming a guest poster here as well, don’t hesitate to contact me. Vincent.

Rob Close on Rob Close: I am a student pilot currently based in Sydney, Australia. I currently hold a GFPT (restricted PPL) and have flown Cessna 152s, Piper Warriors, CT4Bs and the CAP10. My aim is to be a commercial pilot and will be studying a Bachelor of Aviation at Swinburne University in Melbourne this year. My blog (http://theflyinghighblog.blogspot.com/) focuses on my aviation adventures and travels around Australia and the world.

Pictures:
Singularity University Banner: By david.orban
Yemenia Aircrew: By January 152010
Northwest A320: By Caribb

Tagged as: error, hypothesis, neuroscience, pilot error