AN INTERNATIONAL JOURNAL
DEVOTED TO THE SCIENTIFIC STUDY OF HUMAN
PERFORMANCE AND HUMAN FACTORS AT WORK,
MACHINE CONTROL AND EQUIPMENT DESIGN
VOL. 14, No. 5, SEPTEMBER 1971
Variations in ATC-Work Load as a Function of Variations in Cockpit Workload
By Haylitt Retief
Laboratory for Ergonomie Psychology, TNO, Zuiderzeeweg 10. Amsterdam,
Holland
The relation between pilots' workload and radar
controllers' workload was investigated. Pilots' workload served as the independent
variable. It was assumed that a number of levels of workload can be distinguished in a
standard airport traffic circuit under visual flight rules. The dependent variable
(controllers' workload) was measured by response frequency on an auditory binary choice
task.
Results indicate a rise in radar controllers' workload in a predicted direction for a
number of pilots' workload levels.
1. Introduction
In this experiment the dual task method was used in two ways.
(1) Traditionally, measuring the so-called spare mental capacity. In this case the subject
is asked to perform a secondary task at the same time, which serves to absorb the mental
capacity not used by the main task. In this way an indication is obtained of the mental
load implied in the performance of a job. Borneman (1942), Poulton (1950), Brown (1962),
Haider (1962), Michon (1966), Rolfe (1966, 1969) and others have used this method.
(2) The method of distraction stress. In experiments using distraction stress the main
task to which preference has to be given is a simple repetitive one. The latter, for
instance a binary choice task, can be regarded as a stress condition in the performance of
a secondary task. This way of employing the dual task method has been called ' The Method
of Distraction Stress ' (Kalsbeek 1962, 1964 and 1971). In experiments using distraction
stress normal task performance can be systematically broken down by progressively
occupying the subject's channel capacity, e.g. by increasing the rate of binary choices.
Differences in performance which might be less pronounced without distraction stress can
become measurable with distraction stress.
The aim of our experiments has been to investigate the relationship between the pilot's
workload and the ATC controller's workload. This paper, however, deals only with the
following questions. Can the binary choice task be used in actual flight in order to cause
deterioration of flying performance? Could it be used to measure the radar controller's
spare mental capacity?
It is assumed that different levels of a pilot's mental workload can be distinguished in a
simple airport traffic circuit under visual flight rules.
It is predicted that the average response rate of the radar controller to the binary
choice task will decrease as a function of increasing the controller's workload and that
this in turn will increase as a function of the pilot's workload.
2. Method
2.l. Apparatus
The radar controller was presented with two foot pedals. The choice was which of the two
pedals had to be pressed: the left pedal on hearing a low tone and the right one when the
tone was high. The tones were generated according to a random schedule and administered
through a headset. When the subject responded incorrectly this was automatically
registered. The BCG incorporated a panel on which the errors and the total number of
responses made were digitally displayed. In the case of the radar controller, where the
response frequent was the workload measure, the number of tones generated per unit time
was determined by the subject. A new tone was presented only after a response to the
previous one had been made. The tone discrimination task in this case was the secondary
task. In the aircraft the pilot had two push-buttons conveniently situated on the top left
side of the control column. With his left thumb he pressed the left button for a low tone
and the right one for a high tone. On a portable tape recorder a recording was made of the
BCG tones at a thirty tones per minute frequent. In the approaches with distraction stress
the recorder was switched on at the start of the descent. Obviously in this case the
number of tones generated per unit time is independent of the subject's responses. Thus a
tape recording could be -used. Wrong responses as well as omissions to respond were
recorded. Here the tone diserimination task was the primary task.
2.2. Procedure
Six standard circuits were flown (the aircraft was a single-engined Cessna 172). This was
done at Schiphol Airport, starting approximately at midnight. The circuits were flown on
instruments ' under complete radar control. The pilot flew the courses which the radar
controller told him to fly. The approach was of the S.R.A. type, controlled to a height of
500 feet, two miles from the runway threshold. After this point the pilot continued with a
visual approach. He made a touch and go and reverted back to radar control at 100 feet,
from then on flying only on instruments.
Figure 1 shows the circuit flown by the aircraft. The solid parts of the line mark the
sampling periods in which the controller manipulated the BCG. The words between quotation
marks recurred in the radio traffic in each circuit. They served as cues for the ground
experiments to start the BCG. Four levels of workload were assumed to exist: (1) level
flight, (2) take off, (3) approach, (4) approach with distraction stress. The order
supposedly represents increasing stress.
Additionally a BCG response frequent was established when the controller had no other work
to do ('no load'). 'No load' was established after a 10 minutes learning period, some 30
minutes before take-off for the first flight. In the air the experiments (a
pilot-psychologist) was situated just behind the pilot. He could check the pilot's BCG
responses by means of a simple device. Whenever a wrong response or omission was made, the
experimenter illuminated one of four lights on a small box visible to the pilot. The pilot
was briefed to make an immediate overshoot if he made more than three mistakes (i.e. if
the fourth light came on). The necessity of giving preference to the BCG task over flying
the aircraft led to a considerable deterioration in flying performance.
2.3. Recording results
The scores were recorded in two ways. First by direct observation of the BCG digital
response display panel. Additionally a video tape recording was
Figure 1. Circuit flown by experimental aircraft,. The thick black line shows sampling periods.
made of this display panel. The picture thus obtained was mixed
electronically with a video recording of the radar screen during all the sampling periods.
This resulted in a final television picture where the BCG response display appeared above
the radar screen. We thus had a continuous visual (and aural) record of the ATC
controllers' activity with above it a running display of what was assumed to be a measure
of his workload at the time.
Apart from average response per minute values this form of recording gives data which lend
themselves to qualitative analysis of the total periods.
3. Results
Results are shown in Figure 2. The production per minute is displayed on
the vertical axis. The amount increases from top to bottom, to give a rising curve with
increasing load. The upper curve represents the average error score per minute. The load
levels are shown on the horizontal axis.
R is the value obtained with no load at all.
LF is the average value in the sampling periods when the aircraft is in level flight
downwind.
TO is when the aircraft is taking off, with the controller anticipating taking over
control.
Figure 2. Radar controllers' BCG responses at various stages of flight.
4. Conclusions
(1) It appears to be technically possible to use the BCG as a measure of
' spare capacity' at the same time as a control task is carried out.
(2) lt appears possible to administer distraction stress in flight to such an extent that
the approaches will deteriorate.
(3) The results suggest that the workload measure employed shows a trend in the predicted
direction. Going strictly by the results, it must be concluded that the controller is less
'loaded' when he is controlling the aircraft in level flight, than when he is not
controlling it at all. Additionally it makes no difference to the controller if the
pilot's performance is impaired by additional stress.
5. Discussion
Considering the approaches under distraction stress, from a controller's
workload point of view the following must be taken into account. Though these approaches
deteriorated considerably this deterioration was especially marked in glide path
deviation. It was hoped by the experimenter that extended-centerline deviations would also
have been considerable. This would have resulted in an increase in the controller's
workload, since more heading corrections would have been required. With the equipment used
in this experiment (surveillance radar) optimal-glide-path deviations do not lead to an
increase in the controller's activity, since he has only control in azimuth. With
extended-centerline deviations being least affected, it is not surprising that the
deterioration of the pilot's performance, induced by distraction stress, is not reflected
in the controller's workload as measured by his BCG performance.
It might be argued that the ATC situation with one aircraft in the circuit is a very
simple one. Reduction of the field of study is, however, the first step in controlled
research. Reduction of the problem area, without losing too much relevance, is sound
practice at all times. It is only thus that one can arrive at the expression of relevant
variables on scales which are not too presumptions, thereby invalidating one's statistics.
Conditions can be made progressively more complex as firm knowledge is gained.
This applies to the situation described here. Clearly it would have been a simple matter
to have a second aircraft in the circuit at the same time. If this aircraft was made to
fly exactly the same circuit, the situation would already have been more complex, but not
necessarily less controlled from an experimental point of view.
Now if this second aircraft was made to fly at a different speed from the first one, the
situation would again have been more complex, but not necessarily less controlled. One can
easily imagine how one could increase the number of aircraft and vary their speeds to
obtain various degrees of complexity in a controlled design.
Finally a word with regard to the safety factor. Obviously when an aircraft is being
talked down by radar under realistic conditions (the experiment was done at night) and
both pilot and radar controller are being systematically distracted, one has to give this
some thought. In this experiment it was strictly emphasized and formally established that
the radar controller (though a very senior controller himself) was working under the
supervision and responsibility of a second controller. The pilot of the aircraft (though a
professional pilot) was not legally in command. Thus a situation is created which can be
compared with a pupil radar controller talking down a pupil pilot. This is the only
responsible way this type of experiment can be done.
La relation entre la charge de travail des pilotes et cello des
contro1eurs au radar a ete etudiée. La charge de travail des photos a servi de variable
indépendante. On a supposé qu'un certain nombre de niveaux de charge de travail
pouvaient étre distingués sur un circuit de 1'aéroport effectué dans des conditions
normal es de vol. La cadence de reponse dans une tache auditive de choix binaires
constitualt a variable dependante (charge de travail des contr61exirs). Les résultats
montrent un accroissement de la charge de travail des contro1eurs en function des niveaux
de la charge de travail des pilotes, ce qui est conforme aux prédictions.
Die 13 beziehung zwischen der Arbeitsbelastung eines Flugzeugführers und eines Radar Kontrolleurs wurde untersucht. Die Belastung des Flugzeugführers diente als unabhängige Variable. Es wurde angenommen, dass sich eine Anzahl von Belastungsstufen bei einem Standard-Luftverkehrsablauf unter Sicht-bedingungen unterscheiden lassen. Die abhängige Variable, die Belastung des Kontrolleurs, wurde mit einem binären, auditorischen Wahlreaktionstest gemessen. Die Resultate zeigen einen Anstieg der Belastung des Radar-Kontrolleurs in einer vermuteten Richtung für eine Anzahl von Stufen der Flugzeugführer-Belastung.
References
BORNEMAN, E., 1942, Untersuchungen über den Grad der geistigen
Beanspruchung. Arbeitsphysiologie, 12, 142.
HAIDER, M., 1962, Experimentelle Untersuchungen über geistiger Beanspruchung durch
Arbeitsleistungen. Zeitschriftfur Angewandte.Physiologie einschlieszlich Arbeits
Physiologie, 19, 241-251.
KALSBEEK, J. W. H., 1962, Loss of information in information generating processes as a
measure of distraction stress induced by the method of subsidiary tasks. XXII
International Congress of Physiological Scienceg, Excerpta -Aledica, International
Congress Series No. 48, p. 1104.
KALSBEEK, J. W. H., 1964, On the measurement of deterioration in performance caused by
distraction stress. Ergonomics, 7, 187-195.
KALSBEEK, J. W. H., 1971, Sinus arrhythmia and the dual task method in measuring mental
load. In Measurement of Man at Work (Edited by W. T. SINGLETON, J. G. Fox and D.
WHITFIELD) (London: TAYLOR & FRANcis LTD.), 101-113.
RETIEF, H. and OPMEER, C. H. J. M., 1971, The objective scoring of flight-performance,
combined with a distraction task during the approach. To be published.
MICHON, J. A., 1966, Tapping regularity as a measure of perceptual motor load. Ergonomics,
9, 5.
POULTON, E. C., 1958, Measuring the order of difficulty of visual motor tasks. Ergonomics,
1, 234-239.
ROLFE, J. M., 1969, An evaluation of the effectiveness of a secondary task. RAF IAM Report
473.
Discussion
One problem which arose in the discussion was that there were no obvious
differences between the results of the two test conditions: 'normal work-load' and
'work-load with added binary choice generator '. Retief referred to what had been said
about the effect of distraction stress on the pilots flying performance: the effect was
obvious, but in a parameter that did not concern the radar controller with this type of
approach procedure. He admitted that the results of the experiment did not agree with the
hypotheses for all workload levels. 13ut to prove the findings statistically more
experiments must be carried out.
With regard to safety in the experiments Retief explained that the pilot flying the
aircraft was not legally in command. The pilot legally in command was carefully observing
flight progress ready to take over control if deterioration should become unacceptable to
him. The same arrangement applied to the radar controller. Also, the plan was worked out
carefully and discussed with all participants so that there were no objections either from
the legal point of view or from the participants themselves.
return to About Haylitt - return to start page - return to International page
Copyright Haylitt Retief ©