adrenaline & gain

[From Bill Williamls UMKC 26 December 2002 2:50 PM CST]

from Aviation Week & Space Technology December 16 2002 p. 8.

This following is from a discussion of the crash of American Airlines
flight 587.

"I am reminded of a conversation on a flight years ago with a senior
executive of a major aerospace company who had been involved in the
development of gain schedules for some of the first military fly-by-wire
aircraft. He said that despite a lot of testing in simulators, they
found real-world flight test simulations in which the total control
loop (including the pilot) was unstable.
   What they found, he summarized, was that in viewing the human pilot
as a control system element, the effect of adrenaline on the human
system cold be "modeled almost perfectly as a massive increase in gain."
And, he lamented, they cold just about everything to the very experienced
test pilots in the simulator except generate that big hit of pure
adrenaline.

[From Bruce Gregory (2002.12.26.1819)]

What they found, he summarized, was that in viewing the human pilot
as a control system element, the effect of adrenaline on the human
system cold be "modeled almost perfectly as a massive increase in gain."
And, he lamented, they cold just about everything to the very experienced
test pilots in the simulator except generate that big hit of pure
adrenaline.

That certainly matches my experience in recovering from unusual attitudes.
By the way did you see the article in Av. Week also on American 587 that
pointed out that simulators were often programmed in a fashion that encouraged
extreme rudder deflections since the simulation did not reveal the
extreme stresses and potential breakup?

as a control system element, the
effect of adrenaline on the human

system cold be “modeled almost perfectly as a massive increase in
gain.”

And, he lamented, they could [do] just about everything to the very
experienced

test pilots in the simulator except generate that big hit of pure

adrenaline.
From Bill Powers (2002.12.27.0650 MSY)]

Bill Williamls UMKC 26 December 2002 2:50 PM
CST]

What they found, he summarized, was that in viewing the human
pilot

Hard to take a simulation that seriously. The other thing that’s hard to
do in a simulator is generate the changes in pressure on the body due to
accelerations in various directions. Some big simulator installations
actually tip the cockpit this way and that, but then they’re stimulating
the semicircular canals in a way that’s not realistic for linear (or
large-radius) accelerations.

Of course I’d question the assumed temporal sequence: increase adrenaline
–> increase gain. I’d guess it’s more like this:

-----> increase gain

/

/

 Recognize danger

\

\

-----> increase adrenaline

Best,

Bill P.

[From Bruce Gregory (2002.12.27.0927)]

Bill Powers (2002.12.27.0650 MSY)

Of course I’d question the assumed temporal sequence: increase adrenaline
–> increase gain. I’d guess it’s more like this:

                       -----> increase gain

                     /

                   /

  Recognize danger

                   \

                     \

                       -----> increase adrenaline

Your model could be tested by seeing if gain increases as much when adrenaline
release is suppressed.

/

/

 Recognize danger

\

\

-----> increase adrenaline
Your model could be tested by
seeing if gain increases as much when adrenaline release is suppressed.

[From Bill Powers (2002.12.27.0814 MST)]

Bruce Gregory (2002.12.27.0927)–

\
-----> increase gain

Good idea. Is there some way the adrenals could be prevented from
responding to signals from the hypothalamus (or pituitary, I’m pretty
hazy about these systems)? My reasoning is that the effects of adrenaline
are mainly on somatic systems like vasoconstriction, not on the CNS where
gain of a control system would be varied (I assume).

Best,

Bill P.

[From Bill Williams UMKC 27 December 2002 1:30PM CST]

Bruce Gregary 2002.12.26.1819

In regard to your BTW, concerning the programing of simulators so that a
pilot would be encourged to extreme efforts to control by pulsing the
rudder. I didn't understand what they were talking about. THere is in a
following issue of AW a letter arguing that pulsing the rudder is never
justified-- which I think I would agree. But, I've seen people doing this
during twin-engine training when they've made a mistake, actual flight, in
identifying which engine has failed. I think what was going on was a poorly
damped reaction (3-4 overshoots) to the initial mistake. My speculation is
that pulsing the rudder is a behavior that is inadvertantly genrated by a
poorly thought out training program in which an engine failure is
simulated, inflight failure that is, as an abrupt failure-- which is I think
actually quite rare. This combined with an emphasis on making an extremely
quick reaction to an engine failure may be what is sometimes behind the
rudder pulsing. On the other hand perhaps a year ago a 747 with a heavy
fuel load experience a failure of one of the outer engines taking off from
San Francisco. The co-pilot didn't use the rudder at all, or at least very
minimually, and attempted to maintain directional control with the ailrons.
THis very nearly resulted in a loss of control.

I'm sure we will hear more about this as everyone attempts to assign the
blame and legal damages to some other party.

best
bill williams

[From Bruce Abbott (2002.12.27.1705 EST)}

Bruce Gregory (2002.12.27.0927) --

Bill Powers (2002.12.27.0650 MSY)

Of course I'd question the assumed temporal sequence: increase
adrenaline --> increase gain. I'd guess it's more like this:

                          -----> increase gain
                        /
                      /
     Recognize danger
                      \
                        \
                          -----> increase adrenaline

Your model could be tested by seeing if gain >>increases as much when adrenaline release is >>suppressed.

It could be also be tested by injecting adrenalin (epinephrine).

Bruce A.

[From Bill Powers (2002.12.28.1250 MST)]

Bruce Abbott (2002.12.27.1705 EST)}

Bruce G:
>>Your model could be tested by seeing if gain increases as much when
adrenaline >>release is suppressed.

It could be also be tested by injecting adrenalin (epinephrine).

Bruce A.

That's not quite as straightforward. Biochemical changes have effects which
we can perceive, and when we perceive them we tend to interpret them in
terms of the overall state in which they normally occur. So a sudden jolt
of adrenaline injected from outside could be mistaken for fear or anxiety
that normally arise from higher levels in the hierarchy. And that might
well result in an increase in gain!

The basic point in question is whether adrenaline has effects on the CNS
that would increase the ratio of output signal to error signal, which is
the sort of gain change we may be talking about. I suspect not, but I'm no
expert.

Best,

Bill P.

P.S. I'm using Mary's machine after a massive virus attack on my own
machine. 350 files infected, according to Norton. The com ports stopped
working, among other things, and now I can't reinstall Windows ME -- it
gets most of the way through and stops on some error. I have a backup of
sorts on drive D (20 GB from the former computer) but don't know what has
survived there, either.

···

[From Rick Marken (2002.12.28.0800)]

Bill Powers (2002.12.28.1250 MST)--

P.S. I'm using Mary's machine after a massive virus attack on my own
machine. 350 files infected, according to Norton. The com ports stopped
working, among other things, and now I can't reinstall Windows ME -- it
gets most of the way through and stops on some error. I have a backup of
sorts on drive D (20 GB from the former computer) but don't know what has
survived there, either.

That's really terrible. I'm very sorry to hear it. I sure hope you can fix
things up successfully.

But this does make me realize that, while many of your demo programs will not
run on the Mac, neither will most viruses.

Best regards

Rick

···

---
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313