Dear all,
Attached you find a document about HF in aviation. It deals with human performance in the cockpit of an aircraft and why pilots may not execute required procedures. I am very eager to read how you think PCT can help in preventing checklist / procedure omissions.
We discuss this paper in training and I hope that your PCT understanding may lead to new insights for the pilots.
Thank you,
Arthur
Human Factors in aviation.pdf (160 KB)
[From Bill Powers (2009.07.15.1533 MDT)]
Dear all,
Attached you find a document about HF in aviation. It deals with human
performance in the cockpit of an aircraft and why pilots may not execute
required procedures. I am very eager to read how you think PCT can help
in preventing checklist / procedure omissions.We discuss this paper in training and I hope that your PCT understanding
may lead to new insights for the pilots.Thank you,
Arthur
Concerning the article, one idea immediately came to mind: a set of
switches that must be all be manually thrown to enable the throttles to
be advanced to takeoff power. Each switch, of course, goes with a key
element of a checklist and all the switches must be manually operated in
the correct sequence. This assures at least that the pilot or copilot
consciously perceives and focusses attention on the switches, which have
labels that light up (or change from red to green) when thrown. Even a
touch-screen could be used, and the computer program is trivially simple.
For belt-and-suspenders systems, the pilot and copilot could be required
to throw corresponding switches, one set on the pilot’s left and the
other on the copilot’s right – more than an arm’s reach apart.
In problem-solving mode,
Best,
Bill P.
At 09:18 PM 7/15/2009 +0200, Arthur Dijkstra wrote:
LS,
I wonder, why is it that this topic has so little response ?
Is PCT not effective for these kind of problems ?
Is PCT mature for the sort of problems ?
Are the problems not clear or understood ?
Many aviation human factors ‘specialists’ stem from
the psychological school you disagree with. Mostly they are still very
Cartesian and using linear cause effect approach explanations. You can read that
in many accident reports.
I was wondering whether the concept of homeostasis is related to
PCT. The person want to keep his perception within ‘reasonable’
limits of the reference. Also balancing different layers of control with
different references seems like homeostasis to me. The pilots maintaining
control over essential variables within reasonable limits can be seen as
homeostasis I think. I don’t mind a few knots speed difference but not to
much difference, then I have to act to reduce the error by e.g. applying
thrust.
There are several human factor journals that might be willing to
publish PCT human factor articles. Maybe that in the applied science field you
can show more effect than S-R approaches and there is possibly less resistance
to the PCT approach since more and more safety is seen as a control issue.
Best,
Arthur
Van: Control
Systems Group Network (CSGnet) [mailto:CSGNET@LISTSERV.ILLINOIS.EDU] Namens Arthur
Dijkstra
Verzonden: woensdag 15 juli 2009 21:19
Aan: CSGNET@LISTSERV.ILLINOIS.EDU
Onderwerp: PCT in Human Factors
Dear all,
Attached you find a document about HF in aviation. It deals with
human performance in the cockpit of an aircraft and why pilots may not execute required
procedures. I am very eager to read how you think PCT can help in preventing
checklist / procedure omissions.
We discuss this paper in training and I hope that your PCT
understanding may lead to new insights for the pilots.
Thank you,
Arthur
[From Rick Marken (2009.07.22.1150)]
I wonder, why is it that this topic [Human Factors] has so little response ?
I’m probably one of the very few on this list who bills himself as a Human Factors engineer. I didn’t respond to your very interesting post because I thought Bill’s response answered it so nicely. I think you could have gotten more responses if you had responded to Bill’s reply; then discussions might have built on that.
Is PCT not effective for these kind of problems ?
I think PCT can definitely help with there problems. Bill’s analysis was based on PCT, in terms of seeing a possible solution in setting up the situation so that a higher order perception could only be controlled if two people controlled all the appropriate lower level perceptions (and, of course, if the system itself is designed so the environmental connections provided the correct constraints). I’ve used PCT as a framework for determining what are likely to be the main variables contributing to medical error. My approach is described in:
Marken, R. S. (2003) Error In Skilled Performance: A Control Model of Prescribing,
Ergonomics, 46(12), 1200-1214.
I’ve also described how PCT can be used as a basis for task analysis:
Marken, R. S. (1999) PERCOLATe: Perceptual Control Analysis of Tasks, International
Journal of Human-Computer Studies, 50, 481-487
Is PCT mature for the sort of problems ?
Sure. But it’s not the maturity of PCT that matters. It’s the maturity of the people who would apply it to these problems that matters;-)
Are the problems not clear or understood ?
For sure. The problem of error, for example, is terribly confused in Human Factors. PCT can really help out with that.
Many aviation human factors specialists stem from
the psychological school you disagree with. Mostly they are still very
Cartesian and using linear cause effect approach explanations. You can read that
in many accident reports.
Yes, of course. And I’ve found that most (but not all) Human Factors types are just as dismissive of PCT as are most academic (research) psychologists.
I was wondering whether the concept of homeostasis is related to
PCT.
Of course. Homeostasis is just a description of the phenomenon of control with respect to a fixed (as opposed to a variable) reference, as in body temperature control.
The person want to keep his perception within reasonable
limits of the reference. Also balancing different layers of control with
different references seems like homeostasis to me. The pilots maintaining
control over essential variables within reasonable limits can be seen as
homeostasis I think. I dont mind a few knots speed difference but not to
much difference, then I have to act to reduce the error by e.g. applying
thrust.
Yes. It’s all control.
There are several human factor journals that might be willing to
publish PCT human factor articles. Maybe that in the applied science field you
can show more effect than S-R approaches and there is possibly less resistance
to the PCT approach since more and more safety is seen as a control issue.
My experience has been that it’s no more (or less) difficult to publish in Human Factors journals than in any other journals. The problem (for me) has been to get people to 1) pay attention to what’s being presented and 2) understand it.
Best
Rick
Richard S. Marken PhD
rsmarken@gmail.com
Thanks Rick
This is Bills response:
Concerning the article, one idea immediately came to mind: a
set of switches that must be all be manually thrown to enable the throttles to
be advanced to takeoff power. Each switch, of course, goes with a key element
of a checklist and all the switches must be manually operated in the correct
sequence. This assures at least that the pilot or copilot consciously perceives
and focusses attention on the switches, which have labels that light up (or
change from red to green) when thrown. Even a touch-screen could be used, and
the computer program is trivially simple. For belt-and-suspenders systems, the
pilot and copilot could be required to throw corresponding switches, one set on
the pilot’s left and the other on the copilot’s right – more than an arm’s
reach apart.
AD:
A redesign of the cockpit, to enable only thrust advancement when the switches
are in the correct position, is unfortunately beyond the control of the airline
company. The airline has some control over the procedures, Boeing or Airbus has
to agree to the changes, but the airline has more control over the pilot
training. The airline seeks its solutions, to prevent take-off in a wrong
configuration, in a combination of procedures and training. How can PCT help in
those areas ?
Thanks for the name of the publications.
Arthur
Van: Control
Systems Group Network (CSGnet) [mailto:CSGNET@LISTSERV.ILLINOIS.EDU] Namens Richard
Marken
Verzonden: woensdag 22 juli 2009 20:54
Aan: CSGNET@LISTSERV.ILLINOIS.EDU
Onderwerp: Re: PCT in Human Factors
[From Rick Marken
(2009.07.22.1150)]
On Wed, Jul 22, 2009 at 10:40 AM, Arthur Dijkstra Arthur@dijkstraonline.nl wrote:
I wonder, why is it that this
topic [Human Factors] has so little response ?
I’m probably one of the very few on this list who bills
himself as a Human Factors engineer. I didn’t respond to your very interesting
post because I thought Bill’s response answered it so nicely. I think you could
have gotten more responses if you had responded to Bill’s reply; then
discussions might have built on that.
Is PCT not effective for these
kind of problems ?
I think PCT can definitely help with there problems. Bill’s
analysis was based on PCT, in terms of seeing a possible solution in setting up
the situation so that a higher order perception could only be controlled if two
people controlled all the appropriate lower level perceptions (and, of course,
if the system itself is designed so the environmental connections provided the
correct constraints). I’ve used PCT as a framework for determining what are
likely to be the main variables contributing to medical error. My approach is
described in:
Marken, R. S. (2003) Error In Skilled Performance: A Control Model of
Prescribing,
Ergonomics, 46(12), 1200-1214.
I’ve also described how PCT can be used as a basis for task analysis:
Marken, R. S. (1999) PERCOLATe: Perceptual Control Analysis of Tasks,
International
Journal of Human-Computer Studies, 50, 481-487
Is PCT mature for the sort of
problems ?
Sure. But it’s not the maturity of PCT that matters. It’s
the maturity of the people who would apply it to these problems that matters;-)
Are the problems not clear or
understood ?
For sure. The problem of error, for example, is terribly
confused in Human Factors. PCT can really help out with that.
Many aviation human factors
‘specialists’ stem from the psychological school you disagree with.
Mostly they are still very Cartesian and using linear cause effect approach
explanations. You can read that in many accident reports.
Yes, of course. And I’ve found that most (but not all)
Human Factors types are just as dismissive of PCT as are most academic
(research) psychologists.
I was wondering whether the
concept of homeostasis is related to PCT.
Of course. Homeostasis is just
a description of the phenomenon of control with respect to a fixed (as opposed
to a variable) reference, as in body temperature control.
The person want to keep his
perception within ‘reasonable’ limits of the reference. Also
balancing different layers of control with different references seems like
homeostasis to me. The pilots maintaining control over essential variables
within reasonable limits can be seen as homeostasis I think. I don’t mind
a few knots speed difference but not to much difference, then I have to act to
reduce the error by e.g. applying thrust.
Yes. It’s all control.
There are several human factor
journals that might be willing to publish PCT human factor articles. Maybe that
in the applied science field you can show more effect than S-R approaches and
there is possibly less resistance to the PCT approach since more and more
safety is seen as a control issue.
My experience has been that it’s no more (or less) difficult
to publish in Human Factors journals than in any other journals. The problem
(for me) has been to get people to 1) pay attention to what’s being presented
and 2) understand it.
Best
Richard S. Marken PhD
rsmarken@gmail.com
[From Rick Marken (2009.07.22.1330)]
Hi Arthur. It would be nice if you would put a header like mine above on your posts. It’s kind of a CSG convention.
AD:
A redesign of the cockpit, to enable only thrust advancement when the switches
are in the correct position, is unfortunately beyond the control of the airline
company. The airline has some control over the procedures, Boeing or Airbus has
to agree to the changes, but the airline has more control over the pilot
training. The airline seeks its solutions, to prevent take-off in a wrong
configuration, in a combination of procedures and training. How can PCT help in
those areas ?
I don’t know if it can help any more than conventional HF in that situation. One thing I did show in the paper on medical error is that increased training isn’t very cost effective when error rate is already very low. In that situation, you are probably better off focusing on developing good, standardized procedures. And, of course, redundancy always helps so one standard “meta” procedure should probably be to have all procedures carried out by both the pilot and copilot, one after the other.
Best
Rick
Richard S. Marken PhD
rsmarken@gmail.com
[From Bill Powers (2009.07.22.1346 MDT)]
AD: A redesign of the cockpit, to enable only thrust advancement when the switches are in the correct position, is unfortunately beyond the control of the airline company. The airline has some control over the procedures, Boeing or Airbus has to agree to the changes, but the airline has more control over the pilot training. The airline seeks its solutions, to prevent take-off in a wrong configuration, in a combination of procedures and training. How can PCT help in those areas ?
BP: Ah, another constraint: no structural change in the cockpit. All right, since the object is to affect a perception in the pilot, let's think of a cheap gadget that will demand the pilot's attention to prevent the pilot from trying to take off before all (critical) items in the checklist have been completed.
This one requires a mounting bracket or clamp that will support a small device located aft of the position of the throttles (which are pushed forward to take off). This device contains a small computer which manages the checklist and receives checkoff reponses. Plugged into the forward end of this device is a bungee cord, the other end of which is a loop that is passed over the throttle handles. The length and elasticity of the bungee cord are selected so that there is no resistance at all for throttle settings used for taxiing around on the ground, but definitely strong resistance to pushing the throttle to takeoff settings. An elastic cord is used instead of a restraining cable so the pilot can get full power if really needed. The signal to the pilot that the checklist has not been completed is simply the sensation of abnormally large effort that is felt when the bungee cord is still attached to the box. A significant pull on the bungee cord tells the computer that the throttles are being pushed beyond the allowed range. Completing the checklist releases the bungee cord.
Of course a lighter cord can be used that simply pulls out of the box when the throttle is pushed far enough forward, setting off a godawful racket or a verbal message ("YOU DIDN'T DO STEP 7, YOU IDIOT") that the pilot can't ignore.
This requires only a small portable device the size of a pocket calculator that the pilot can fasten in place and remove. Putting it in place is the first step in the takeoff ritual.
Wait a minute -- there's another way to detect the application of takeoff power, courtesy of Isaac Newton. Who cares if the throttles are pushed forward? They don't make the airplane take off. The engines do. All we need is an accelerometer in the little checkoff box that the pilot uses.
Analog Devices makes one that detects all three axes -- the pilot could leave the little box in his pocket, except for entering checkoffs. The pilot and copilot could have boxes that talk to each other. Would it be OK to have the earphone cables plug into the box (and the box into the radio for normal use) so the box could talk to the pilot that way? It could actually read the checkoff list to him.
Of course detection of takeoff acceleration would immediately produce a very urgent audio signal if the checklist has not been completed.
Here's a distributor: less than $10 in single quantities.
http://www.sparkfun.com/commerce/product_info.php?products_id=9265
Now you don't need any modifications at all in the cockpit.
Am I hired?
Best,
Bill P.
At 09:17 PM 7/22/2009 +0200, Arthur Dijkstra wrote:
from Arthur Dijkstra 2009-07-23
Thanks Bill,
Yes attention is crucial, but attention must be paid to many issues. Your
engineering solution is interesting but has to be approved and certified if
any connection with the aircraft is made or when transmitters are used.
Aircraft manufacturers have installed a take-off configuration warning which
should given an enormous sound and warning light when the configuration is
not correct and thrust lever are advanced. It is airlines policy however
that for crucial (live or die) events we don't want to rely (to much) on a
last line of defence. In some take-off accidents the take-off warning was
not working. Anyway that is some background and drifting us away from PCT.
I will forward your application to our flight technical department...
Best,
Arthur
[From Bill Powers (2009.07.22.1346 MDT)]
AD: A redesign of the cockpit, to enable only thrust advancement
when the switches are in the correct position, is unfortunately
beyond the control of the airline company. The airline has some
control over the procedures, Boeing or Airbus has to agree to the
changes, but the airline has more control over the pilot training.
The airline seeks its solutions, to prevent take-off in a wrong
configuration, in a combination of procedures and training. How can
PCT help in those areas ?
BP: Ah, another constraint: no structural change in the cockpit. All
right, since the object is to affect a perception in the pilot, let's
think of a cheap gadget that will demand the pilot's attention to
prevent the pilot from trying to take off before all (critical) items
in the checklist have been completed.
This one requires a mounting bracket or clamp that will support a
small device located aft of the position of the throttles (which are
pushed forward to take off). This device contains a small computer
which manages the checklist and receives checkoff reponses. Plugged
into the forward end of this device is a bungee cord, the other end
of which is a loop that is passed over the throttle handles. The
length and elasticity of the bungee cord are selected so that there
is no resistance at all for throttle settings used for taxiing around
on the ground, but definitely strong resistance to pushing the
throttle to takeoff settings. An elastic cord is used instead of a
restraining cable so the pilot can get full power if really needed.
The signal to the pilot that the checklist has not been completed is
simply the sensation of abnormally large effort that is felt when the
bungee cord is still attached to the box. A significant pull on the
bungee cord tells the computer that the throttles are being pushed
beyond the allowed range. Completing the checklist releases the bungee cord.
Of course a lighter cord can be used that simply pulls out of the box
when the throttle is pushed far enough forward, setting off a
godawful racket or a verbal message ("YOU DIDN'T DO STEP 7, YOU
IDIOT") that the pilot can't ignore.
This requires only a small portable device the size of a pocket
calculator that the pilot can fasten in place and remove. Putting it
in place is the first step in the takeoff ritual.
Wait a minute -- there's another way to detect the application of
takeoff power, courtesy of Isaac Newton. Who cares if the throttles
are pushed forward? They don't make the airplane take off. The
engines do. All we need is an accelerometer in the little checkoff
box that the pilot uses.
Analog Devices makes one that detects all three axes -- the pilot
could leave the little box in his pocket, except for entering
checkoffs. The pilot and copilot could have boxes that talk to each
other. Would it be OK to have the earphone cables plug into the box
(and the box into the radio for normal use) so the box could talk to
the pilot that way? It could actually read the checkoff list to him.
Of course detection of takeoff acceleration would immediately produce
a very urgent audio signal if the checklist has not been completed.
http://www.analog.com/en/sensors/inertial-sensors/adxl330/products/product.h
tml
Here's a distributor: less than $10 in single quantities.
http://www.sparkfun.com/commerce/product_info.php?products_id=9265
Now you don't need any modifications at all in the cockpit.
Am I hired?
Best,
Bill P.
At 09:17 PM 7/22/2009 +0200, Arthur Dijkstra wrote:
[From Bill Powers(2009.07.23.0710 MDT)]
Arthur Dijkstra 2009-07-23
AD: Yes attention is crucial, but attention must be paid to many issues. Your
engineering solution is interesting but has to be approved and certified if
any connection with the aircraft is made or when transmitters are used.
My last idea uses a self-contained inertial accelerometer in a hand-held device that has no connection to aircraft structure or electronics. It doesn't emit anything but light (the display) and sound (the warning). The option for warning the pilot through the headphones would involve an audio cable; unplug the headset, plug the handheld box cable in where the headset was, plug the headset into the handheld box. Would that count as a "connection with the aircraft?"
Anyway that is some background and drifting us away from PCT.
I will forward your application to our flight technical department.
Excellent. Just don't tell them how old I am.
I was looking for a relatively foolproof solution that doesn't depend on pilot training. The best solution would simply make it impossible to get takeoff power if anything important was not checked off. Then training wouldn't matter much and neither would distractions. But if we can't use aircraft modifications, we have to look at working with the pilot.
Distractions have the effect of dragging attention away from control tasks, leaving unattended control systems to work automatically (like breathing). Automatic control systems can't adapt to new situations (can't be reorganized -- according to current PCT assumptions). The basic problem comes down to finding a way to demand pilot attention without creating hazards when there is a need for attention elsewhere -- avoiding a taxiway collision, for example. Using the inertial accelerometer gives us a way to detect when takeoff is attempted without needing any connection to the airplane at all, mechanical or electronic. So the remaining problem (if the checklist is incomplete) is how to get the pilot or copilot's attention without forcibly overriding it, so emergencies can still be dealt with (you wouldn't want to get it by dropping a little curtain in front of the pilot's eyes, say out of his hat).
I'm sure the conventional approach would be cast in terms of conditioning: training the pilot to respond to some special stimulus in the manner of classical conditioning or operant conditioning. Even if we believed those theories, however, it would not be a good idea to give a stimulus the power to cause an unconscious response beyond the pilot's control, because that could happen just at the wrong time and cause an accident. The best solution is to somehow get the pilot's attention without interfering with the pilot's ability to ignore the checklist problem to handle a bigger and more immediate problem.
So what does that leave us? Loud sounds, recorded human voices, vibrators, shocks, and ... ?
We could required the pilot or copilot to hit a button to turn the signal off. The requirement for some action to turn it off would help focus attention in the right place. Attention goes to systems experiencing error. By detecting actual takeoff acceleration, we can be pretty sure that most emergency-type actions would have been dealt with before getting to that point. The aircraft would still be moving slowly enough to abort the takeoff safely -- plenty of runway ahead. I think an audible signal would be best -- the copilot could hear it, too.
Ah, well, I guess I'm not destined for a new career just now.
Best,
Bill P.
[From Arthur Dijkstra 2009.07.23]
Thanks Bill.
I will leave this for now since I am packing for my holiday to the US. The
PCT books will be in my luggage.
Best,
Arthur
[From Bill Powers(2009.07.23.0710 MDT)]
Arthur Dijkstra 2009-07-23
AD: Yes attention is crucial, but attention must be paid to many issues.
Your
engineering solution is interesting but has to be approved and certified if
any connection with the aircraft is made or when transmitters are used.
My last idea uses a self-contained inertial accelerometer in a
hand-held device that has no connection to aircraft structure or
electronics. It doesn't emit anything but light (the display) and
sound (the warning). The option for warning the pilot through the
headphones would involve an audio cable; unplug the headset, plug the
handheld box cable in where the headset was, plug the headset into
the handheld box. Would that count as a "connection with the aircraft?"
Anyway that is some background and drifting us away from PCT.
I will forward your application to our flight technical department.
Excellent. Just don't tell them how old I am.
I was looking for a relatively foolproof solution that doesn't depend
on pilot training. The best solution would simply make it impossible
to get takeoff power if anything important was not checked off. Then
training wouldn't matter much and neither would distractions. But if
we can't use aircraft modifications, we have to look at working with the
pilot.
Distractions have the effect of dragging attention away from control
tasks, leaving unattended control systems to work automatically (like
breathing). Automatic control systems can't adapt to new situations
(can't be reorganized -- according to current PCT assumptions). The
basic problem comes down to finding a way to demand pilot attention
without creating hazards when there is a need for attention elsewhere
-- avoiding a taxiway collision, for example. Using the inertial
accelerometer gives us a way to detect when takeoff is attempted
without needing any connection to the airplane at all, mechanical or
electronic. So the remaining problem (if the checklist is incomplete)
is how to get the pilot or copilot's attention without forcibly
overriding it, so emergencies can still be dealt with (you wouldn't
want to get it by dropping a little curtain in front of the pilot's
eyes, say out of his hat).
I'm sure the conventional approach would be cast in terms of
conditioning: training the pilot to respond to some special stimulus
in the manner of classical conditioning or operant conditioning. Even
if we believed those theories, however, it would not be a good idea
to give a stimulus the power to cause an unconscious response beyond
the pilot's control, because that could happen just at the wrong time
and cause an accident. The best solution is to somehow get the
pilot's attention without interfering with the pilot's ability to
ignore the checklist problem to handle a bigger and more immediate problem.
So what does that leave us? Loud sounds, recorded human voices,
vibrators, shocks, and ... ?
We could required the pilot or copilot to hit a button to turn the
signal off. The requirement for some action to turn it off would help
focus attention in the right place. Attention goes to systems
experiencing error. By detecting actual takeoff acceleration, we can
be pretty sure that most emergency-type actions would have been dealt
with before getting to that point. The aircraft would still be moving
slowly enough to abort the takeoff safely -- plenty of runway ahead.
I think an audible signal would be best -- the copilot could hear it, too.
Ah, well, I guess I'm not destined for a new career just now.
Best,
Bill P.
[From Arthur Dijkstra 16-08-09 09:09utc]
AD: While doing human factor research a PCT seed has been planted in me and
I evaluate and try PCT application. I came across the document mentioned
below and like to know:
Do you agree with the usage and application of PCT in this paper ?
http://pubs.drdc.gc.ca/PDFS/zbb55/p506682.pdf
If not how is PCT misinterpreted ?
Thanks
Arthur
[From Bill Powers (2009.08.16.0832 MDT)]
Arthur Dijkstra 16-08-09 09:09
AD: Do you agree with the usage and application of PCT in this paper ?
http://pubs.drdc.gc.ca/PDFS/zbb55/p506682.pdf
If not how is PCT misinterpreted ?
Hendy is a former colleague of Martin Taylor at DCIEM in Canada, so we can presume that he has more than a passing knowledge of PCT. When I read papers like this, where PCT is embedded in a system of thought thats uses entirely different terms and meanings, I mainly look for what is right, and there is a lot of that. Where non-pct concepts are used, it's harder to tell how they relate to PCT.
I'm not sure that the "mental model" really belongs in the output function; I should think it would be associated with the input function since one can perceive it. The imagination mode is probably involved. But I'd like to see more explorations in which actual mental models are used, to get a better idea of what is happening. I don't think mental models are as central to behavior as Hendy seems to think they are, but I don't know exactly what he thinks a mental model is -- is he really referring to the perceptual hierarchy?
Hendy mentioned my statement that behavior does not reliably repeat when tasks are performed, but proposed that maybe in some situations specific behaviors can be used to achieve specific results. Here the problem is in distinguishing the results of behavior from the actions that produce the results. If you define a behavior as "opening the door", it might seem that you generate the same output every time you go through a door -- but "opening the door" is not a behavior. The door becomes open because you twist the knob and push it open, or if you're going through the door the other way, pull it open. Exactly opposite actions to create "the same behavior." It's really very seldom that we generate the same outputs to create the same behavioral result.
Best,
Bill P.
[From Rick Marken (2009.08.16.1420)]
Bill Powers (2009.08.16.0832 MDT)
Arthur Dijkstra (16-08-09 09:09)
AD: Do you agree with the usage and application of PCT in this paper ?
http://pubs.drdc.gc.ca/PDFS/zbb55/p506682.pdf
If not how is PCT misinterpreted ?BP: Hendy is a former colleague of Martin Taylor at DCIEM in Canada, so we can
presume that he has more than a passing knowledge of PCT. When I read papers
like this, where PCT is embedded in a system of thought thats uses entirely
different terms and meanings, I mainly look for what is right, and there is
a lot of that.
I didn't see all _that_ much that was right with it. They did get the
control theory diagram right. And they did correctly note that the
system acts to keep perceptual representations of the environment in
goal (reference) states. But I don't think it's really right to say,
as they did, that their matrix representation of the functions and
vector representation of the variables is a representation of the
"hierarchy of control". There is no hierarchy shown, or used in the
analysis. I think they also inappropriately superimpose ideas from the
IT model onto the control model. For example, they say that transport
delays in the PCT model could be said to depend on the amount of
information that has to be processed in going sensory input to
perception and from error signal to output. Ignoring that fact that
this implies that information is actually something that control
systems "process", I think PCT would say these transport lags have
more to do with the level (in the control hierarchy) of the perceptual
variable under control (see my
http://www.mindreadings.com/ControlDemo/HP.html for a demonstration of
this phenomenon) rather than with the amount of information being
processed.
My main problem with the paper, though, is that they don't really
_use_ the PCT model for much other than to say that a portion of the
control loop could be called a "mental model". That is, the model is
used as a metaphor rather than as a working model of an observed
behavior. As a human factors engineer myself I wouldn't know what I
had learned from this paper about the usefulness of the PCT model. The
only data presented is a factor analysis of a couple of tests. This
analysis is kind of thrown in at the end, apparently so that there
would be some data to show. But I don't know how the analysis is
related to PCT or what I would do with the analysis based on the
observed results.
I may be biased but I much prefer the three papers I wrote that apply
PCT to human factors issues:
Marken, R. S. (2005) A Model-Based Approach to Prioritizing Medical
Safety Practice, in K. Henriksen, J. Battles, E. Marks & D. Lewin
(Eds) Advances in patient safety: From research to implementation,
Vol. 2, Concepts and methodology. AHRQ Publication No. 05-0-21-2.
Rockville, MD: Agency for Healthcare Research and Quality, 409-424
Marken, R. S. (2003) Error In Skilled Performance: A Control Model of
Prescribing, Ergonomics, 46(12), 1200-1214.
Marken, R. S. (1999) PERCOLATe: Perceptual Control Analysis Of Tasks,
International Journal of Human-Computer Studies, 50, 481-487
I'd be interested in hearing Arthur Dijkstra's impression of any or
all of these papers.
Thanks
Best
Rick
--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com
Thanks Rick,
AD: I need some time to give you my evaluation. For the time being I must
focus on my thesis and can unfortunately not engage much in somewhat
off-topic issues.
One short reaction is that I think human error is a problematic concept.
This concept is an important part of the argument in your paper on faulty
medication to bring your PCT views to the front. The 'error' problem makes
the PCT arguments less relevant for me (in aviation). Furthermore I think,
human error is more of a component failure that a systemic failure and that
limits your interventions. Safety science, at least the school I am part
off, has moved to a systemic approach.
When human error is taken a unproblematic your examples of PCT are
impressive.
Back to my thesis,
Thanks
Arthur
-----Oorspronkelijk bericht-----
Van: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] Namens Richard Marken
Verzonden: zondag 16 augustus 2009 23:18
Aan: CSGNET@LISTSERV.ILLINOIS.EDU
Onderwerp: Re: PCT in Human Factors
[From Rick Marken (2009.08.16.1420)]
Bill Powers (2009.08.16.0832 MDT)
Arthur Dijkstra (16-08-09 09:09)
AD: Do you agree with the usage and application of PCT in this paper ?
http://pubs.drdc.gc.ca/PDFS/zbb55/p506682.pdf
If not how is PCT misinterpreted ?BP: Hendy is a former colleague of Martin Taylor at DCIEM in Canada, so we
can
presume that he has more than a passing knowledge of PCT. When I read
papers
like this, where PCT is embedded in a system of thought thats uses
entirely
different terms and meanings, I mainly look for what is right, and there
is
a lot of that.
I didn't see all _that_ much that was right with it. They did get the
control theory diagram right. And they did correctly note that the
system acts to keep perceptual representations of the environment in
goal (reference) states. But I don't think it's really right to say,
as they did, that their matrix representation of the functions and
vector representation of the variables is a representation of the
"hierarchy of control". There is no hierarchy shown, or used in the
analysis. I think they also inappropriately superimpose ideas from the
IT model onto the control model. For example, they say that transport
delays in the PCT model could be said to depend on the amount of
information that has to be processed in going sensory input to
perception and from error signal to output. Ignoring that fact that
this implies that information is actually something that control
systems "process", I think PCT would say these transport lags have
more to do with the level (in the control hierarchy) of the perceptual
variable under control (see my
http://www.mindreadings.com/ControlDemo/HP.html for a demonstration of
this phenomenon) rather than with the amount of information being
processed.
My main problem with the paper, though, is that they don't really
_use_ the PCT model for much other than to say that a portion of the
control loop could be called a "mental model". That is, the model is
used as a metaphor rather than as a working model of an observed
behavior. As a human factors engineer myself I wouldn't know what I
had learned from this paper about the usefulness of the PCT model. The
only data presented is a factor analysis of a couple of tests. This
analysis is kind of thrown in at the end, apparently so that there
would be some data to show. But I don't know how the analysis is
related to PCT or what I would do with the analysis based on the
observed results.
I may be biased but I much prefer the three papers I wrote that apply
PCT to human factors issues:
Marken, R. S. (2005) A Model-Based Approach to Prioritizing Medical
Safety Practice, in K. Henriksen, J. Battles, E. Marks & D. Lewin
(Eds) Advances in patient safety: From research to implementation,
Vol. 2, Concepts and methodology. AHRQ Publication No. 05-0-21-2.
Rockville, MD: Agency for Healthcare Research and Quality, 409-424
Marken, R. S. (2003) Error In Skilled Performance: A Control Model of
Prescribing, Ergonomics, 46(12), 1200-1214.
Marken, R. S. (1999) PERCOLATe: Perceptual Control Analysis Of Tasks,
International Journal of Human-Computer Studies, 50, 481-487
I'd be interested in hearing Arthur Dijkstra's impression of any or
all of these papers.
Thanks
Best
Rick
--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com
[From Rick Marken (2009.08.26.1050)–
Thanks Rick,
AD: I need some time to give you my evaluation.
And thank you for responding. I don’t quite understand what you said but you are a pilot and I hero worship pilots so I’ll assume you are right;-)
Best
Rick
On Wed, Aug 26, 2009 at 9:28 AM, Arthur Dijkstra Arthur@dijkstraonline.nl wrote:
For the time being I must
focus on my thesis and can unfortunately not engage much in somewhat
off-topic issues.
One short reaction is that I think human error is a problematic concept.
This concept is an important part of the argument in your paper on faulty
medication to bring your PCT views to the front. The ‘error’ problem makes
the PCT arguments less relevant for me (in aviation). Furthermore I think,
human error is more of a component failure that a systemic failure and that
limits your interventions. Safety science, at least the school I am part
off, has moved to a systemic approach.
When human error is taken a unproblematic your examples of PCT are
impressive.
Back to my thesis,
Thanks
Arthur
-----Oorspronkelijk bericht-----
Van: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] Namens Richard Marken
Verzonden: zondag 16 augustus 2009 23:18
Aan: CSGNET@LISTSERV.ILLINOIS.EDU
Onderwerp: Re: PCT in Human Factors
[From Rick Marken (2009.08.16.1420)]
Bill Powers (2009.08.16.0832 MDT)
Arthur Dijkstra (16-08-09 09:09)
AD: Do you agree with the usage and application of PCT in this paper ?
If not how is PCT misinterpreted ?
BP: Hendy is a former colleague of Martin Taylor at DCIEM in Canada, so we
can
presume that he has more than a passing knowledge of PCT. When I read
papers
like this, where PCT is embedded in a system of thought thats uses
entirely
different terms and meanings, I mainly look for what is right, and there
is
a lot of that.
I didn’t see all that much that was right with it. They did get the
control theory diagram right. And they did correctly note that the
system acts to keep perceptual representations of the environment in
goal (reference) states. But I don’t think it’s really right to say,
as they did, that their matrix representation of the functions and
vector representation of the variables is a representation of the
“hierarchy of control”. There is no hierarchy shown, or used in the
analysis. I think they also inappropriately superimpose ideas from the
IT model onto the control model. For example, they say that transport
delays in the PCT model could be said to depend on the amount of
information that has to be processed in going sensory input to
perception and from error signal to output. Ignoring that fact that
this implies that information is actually something that control
systems “process”, I think PCT would say these transport lags have
more to do with the level (in the control hierarchy) of the perceptual
variable under control (see my
http://www.mindreadings.com/ControlDemo/HP.html for a demonstration of
this phenomenon) rather than with the amount of information being
processed.
My main problem with the paper, though, is that they don’t really
use the PCT model for much other than to say that a portion of the
control loop could be called a “mental model”. That is, the model is
used as a metaphor rather than as a working model of an observed
behavior. As a human factors engineer myself I wouldn’t know what I
had learned from this paper about the usefulness of the PCT model. The
only data presented is a factor analysis of a couple of tests. This
analysis is kind of thrown in at the end, apparently so that there
would be some data to show. But I don’t know how the analysis is
related to PCT or what I would do with the analysis based on the
observed results.
I may be biased but I much prefer the three papers I wrote that apply
PCT to human factors issues:
Marken, R. S. (2005) A Model-Based Approach to Prioritizing Medical
Safety Practice, in K. Henriksen, J. Battles, E. Marks & D. Lewin
(Eds) Advances in patient safety: From research to implementation,
Vol. 2, Concepts and methodology. AHRQ Publication No. 05-0-21-2.
Rockville, MD: Agency for Healthcare Research and Quality, 409-424
Marken, R. S. (2003) Error In Skilled Performance: A Control Model of
Prescribing, Ergonomics, 46(12), 1200-1214.
Marken, R. S. (1999) PERCOLATe: Perceptual Control Analysis Of Tasks,
International Journal of Human-Computer Studies, 50, 481-487
I’d be interested in hearing Arthur Dijkstra’s impression of any or
all of these papers.
Thanks
Best
Rick
–
Richard S. Marken PhD
–
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com
[Martin Taylor 2009.08.26.14.27]
[From Arthur Dijkstra 16-08-09 09:09utc]
AD: While doing human factor research a PCT seed has been planted in me and
I evaluate and try PCT application. I came across the document mentioned
below and like to know:
Do you agree with the usage and application of PCT in this paper ?
http://pubs.drdc.gc.ca/PDFS/zbb55/p506682.pdf
If not how is PCT misinterpreted ?
Since you have Keith Hendy's AGARD paper, I thought you might like one of mine, presented at an AGARD conference on "Advanced Aircraft Interfaces: ..." in 1992. It really is two papers, one for what I intended to talk about (the main body) and one for what I did talk about (the last page and a half). I think the latter is the more interesting, because it has more general application to situations in which human and automated systems might work together rather than in alternation.
As totally unbiased critic (of course), I claim that my Layered Protocol Theory is the appropriate specialization of PCT for the analysis and design of human-machine interactions. That's essentially the claim made in this paper. It's 3 MB, and I hope that doesn't cause a problem for the mail system and that it's worth the effort to read.
Martin
PS. I'll be away for the next two weeks, but may have some internet access.
[Martin Taylor 2009.08.26.14.48]
[Martin Taylor 2009.08.26.14.27]
[From Arthur Dijkstra 16-08-09 09:09utc]
AD: While doing human factor research a PCT seed has been planted in me and
I evaluate and try PCT application. I came across the document mentioned
below and like to know:
Do you agree with the usage and application of PCT in this paper ?
http://pubs.drdc.gc.ca/PDFS/zbb55/p506682.pdf
If not how is PCT misinterpreted ?Since you have Keith Hendy's AGARD paper, I thought you might like one of mine, presented at an AGARD conference on "Advanced Aircraft Interfaces: ..." in 1992. It really is two papers, one for what I intended to talk about (the main body) and one for what I did talk about (the last page and a half). I think the latter is the more interesting, because it has more general application to situations in which human and automated systems might work together rather than in alternation.
As totally unbiased critic (of course), I claim that my Layered Protocol Theory is the appropriate specialization of PCT for the analysis and design of human-machine interactions. That's essentially the claim made in this paper. It's 3 MB, and I hope that doesn't cause a problem for the mail system and that it's worth the effort to read.
Martin
PS. I'll be away for the next two weeks, but may have some internet access.
The mailer rejected the message with the attachment because it exceeded the permitted size. I have uploaded it to ftp://ftp.mmtaylor.net/CSG/AGARD1992002.pdf. The login is mmt_ftp, and the password is Anonymous.
Martin