Distributed Cognition in an Airline Cockpit and Airbus Flight Control Design

From this paper: http://hci.ucsd.edu/media/uploads/hci_papers/EH1996-1.pdf

There is a trend in current cockpit design to build two separate crew work
stations for the two pilots. Mechanically linked control yokes are being replaced
in some cockpits by side-stick controllers that are mounted outboard of the
pilot’s seats and are not mechanically linked to each other. From the perspective
of individual pilot performance, side-stick controllers are functionally equivalent
to (or perhaps superior to) control yokes. From the distributed cognition
perspective, however, the side-stick equipped cockpit has a different
distribution of access to information and this may affect the cognitive properties
of the cockpit system.

That was written Edwin Hutchins and Tove Klausen in 1995.

Air France Flight 447 went down 14 years later.

From the Wikipedia article:

In a July 2012 CBS report, Sullenberger suggested the design of the Airbus cockpit might have been a factor in the accident. The flight controls are not mechanically linked between the two pilot seats, and Robert, the left-seat pilot who believed he had taken over control of the aircraft, was not aware that Bonin continued to hold the stick back, which overrode Robert’s own control.

Edwin Hutchins is also the author of Cognition in the Wild which is the subject of another thread here.

not being mechanically linked is one thing but you could eliminate that with servos but to allow the computer to average the inputs is crazy, add that to pilots with poor skill sets and splash…

Not being linked is one thing but it’s the one thing this thread is about. Huchens’ analysis is not in hindsight, it was written several years before the incident. He anticipated the weakness in that set-up.

I’ve changed the title to better reflect the topic.

1 Like

Yes I got that and he was right, Boeing also agreed with him and never went down that route.
There must be hundreds of grey haired simulator instructors that would of said these new pilots are going to fly perfectly serviceable aircraft into the ground or when something goes wrong will crash it when they could of landed.
They were all right if you look at crashes in the last 20 years.

Airline management got same disease as shipping now has, we just need the perfect checklist and nothing can go wrong…

This is from the link in the OP:

In other portions of the flight, especially when the crew is faced with
an equipment failure, the S/O takes other actions that are not visible to the pilots
and notifies them of what he has done. This is only to say that the S/O makes
decisions about the distribution of access to information and organizes his verbal behavior to compensate for the fact that some of his actions are not available to the other members of the crew**

In other words if one of the crew is doing something the others cannot see they would just tell the other crew members. So in the case of Flt AF 447 for example that expectation was not met when the 3/o did not tell the others what he was doing and with the non-linked controls they could not see.

In the wheel house I can’t think off hand of any consequential thing an officer can do without being seen by others in the wheelhouse.

This is from the same paper:

The movement of information through the system has consequences for the
formation of expectations and models of the situation of the aircraft. These
expectations and models organize the behavior of the crew and, when shared,
permit the crew members to coordinate their actions with each other.
Furthermore, the movement of information among members of the crew

1 Like

the difference is a vessel only has one set of controls

Incorrect powerabout. I cannot speak for your vessel, we had 4 sets of controls on my last vessel… Two sets in the lower house, port and starboard, One in the uppper house, and one on the stern controls. You did have to switch air to each station for the throttles. Another switch gave you steering.

geez, and on your boat all 4 can be in use at the same time?

PS you forgot the DP remote at the end of the cable so it can be anywhere and
the Emergency steering in the steering gear room and the engine controls in the engine room.

Reading that report its what should go on in the cockpit and does in well managed operations.
When we get to hear the CVR playbacks from crashes its never like that…

I remember a trainer in a marine schooling the UK i was at updating something, where they also have a full mission sim telling me about sim training there.
Many shipping companies send captains there due to crew complaints and the Captains dont know this.
They jump in the sim ( the crew are the instructors, all current and past Captains) and he said in 100% of the cases they know why the Captain has been sent there in 5 minutes.

Not only that, but there are certain parallels to the McCain accident. That was less about confusion over who was doing what, and more about the station select button being buried in a goddamn menu tree, though. In that respect, this is analogous to the lookout being able to issue helm orders unknown to the conning officer. Quite an HMI design decision, indeed.

EDIT:
This made me curious how things are handled on the F16B. The F16 stick is pressure sensitive, and early iterations didn’t move at all (although there was some give introduced on later models due to PIO), so there is no question of mechanical linkage. Luckily, my mom has hundreds of hours of backseat time in the F16, so she was able to tell me that there is an override switch on the front seat stick. So simple… However, this might work very differently in a training environment, where the expectation is that the trainee doesn’t know what he’s doing, as opposed to a situation where two people are cooperating about flying the aircraft.

2 Likes

Class has lots to do with this.
There are some very dangerous system out there now ( in the DP world) silly thruster companies think they run the vessel and class has allowed them to do processing.

I was on a vessel a couple of years ago assisting in a DP survey and if the thruster didnt receive a command in 40 seconds the thruster control system would take the thruster back and unless you scrolled thru the silly thruster control screen to see that, there was no warning as the ready signal still existed.
The DP cant know its gone till it gives it a command and you get a feedback error.
Only way to get the thruster back was go out of DP and start again.
( This wasnt a one off it was a production system)

Golden rule, never let a thruster company sell you a control system, always use the DP company to make a control system for it.

Did I mention Ethernet on control systems, just as dumb…

On multiple cognitive failures…
Simple switch from manual to DP but switch didnt quite make it to DP so control nowhere. ( class failure, no lights confirming the switch)

Ends up whole crew on the bridge engaged in the perfect tunnel vision error of this cant be happening.
Most people cant shift the thinking to “ok assume it has happened what do we do”

Nobody rechecked the switch they just looked at it
Nobody sent to engine room to get engine control
Nobody looking out the window
20min later drifted into the rig…

I always ask in the DP classes what do you do if the switch fails…
About 2% can give you an answer

1 Like

Thanks for reminding me, there were 5 stations if including emergency steering that ran off the tow winch motor. Did not have DP. Had a handheld remote steering in the lower house along with the mounted sperry units. And no, you could only use one station at a time except lower wheelhouse, they were tied in together. Air had to be shifted to the other throttles you were using.

The McCain is a good example of failure with information processing. There was also an issue between after steering and the bridge but I don’t recall the exact details.

I don’t want to get into specific incidents on this thread but the Exxon Valdez and the El Faro incidents are both good candidates for using this framework for a closer look at how the crew processed information.

2 Likes

You contradicted yourself saying a vessel has only one set of controls, then corrected me about the additional controls I had available. Of course the engine room can take over the power, was referring to access I had on the stations available to the bridge for maneuvering. It wasn’t just one as you stated earlier.

This thread is only indirectly about what crews do. The issue is the first level, not the second.

From the linked paper in the OP

The analysis reveals a pattern of cooperation and coordination of actions
among the crew which on one level can be seen as a structure for propagating
> and processing information and on another level appears as a system of activity
in which shared cognition emerges as a system level property

Seen that way the Airbus design where the least experienced crew member can take control of the aircraft without the other crew being aware seems like a very bad design.

By contrast the paper discusses the traditional layout of the throttles and the origin of the expression “balls to the wall”. The setting of the throttles can be seen by both crew but not the flight controls.

I can say the non-linked control idea is the worst thing to come along since screen doors for submarines. I have never had the displeasure of dealing with trying to fight an Airbus computer into submission, but I was subject to an airplane with electric trim that only worked from one side or the other and it was a nonstop annoyance when the trim “malfunctioned” because the copilot switched it over to his side when I wasn’t looking or I did see it and forgot about it and so on.
On a boat, can you imagine having bow thruster controls or autopilot controls all over the place that did not make it obvious what the other person(s) were doing?
One incident that comes to mind was a sport fisherman with an aft “docking helm station” that had no instruments. Someone at the main helm whacked the autopilot on that promptly overrode any rudder input from me at the aft wheel and I almost ran over a mooring before I realized what was going on, I had no indication at all where I was that Otto had grabbed control.

  • Back to airplanes, Boeing, at least until recently, believed more in the primacy of humans over computers and Airbus was more the opposite. They wanted to sell to 3rd world countries who didn’t trust their flight crews that much from what I read on their philosophy.
1 Like

I was pointing out the difference from aircraft to boat.
I should of said one at a time…

All good.

I often asked my engineers what they would do if the automation failed. Many had no idea and that was understandable. So, we had drills when the company allowed. Shut down critical UPS systems to see if the lads could maintain control. It was a great learning experience for all of us but sadly the companies started forbidding such training.

1 Like