20
This article was downloaded by: [Universite Laval] On: 04 October 2014, At: 07:29 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK The International Journal of Aviation Psychology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/hiap20 Recording of Psychophysiological Data During Aerobatic Training Nicklas Dahlstrom a , Staffan Nahlinder b , Glenn F. Wilson c & Erland Svensson b a Lund University School of Aviation , Ljungbyhed, Sweden b Swedish Defence Research Agency , Stockholm, Sweden c Air Force Research Laboratory , Wright-Patterson Air Force Base , Ohio, USA Published online: 31 Mar 2011. To cite this article: Nicklas Dahlstrom , Staffan Nahlinder , Glenn F. Wilson & Erland Svensson (2011) Recording of Psychophysiological Data During Aerobatic Training, The International Journal of Aviation Psychology, 21:2, 105-122, DOI: 10.1080/10508414.2011.556443 To link to this article: http://dx.doi.org/10.1080/10508414.2011.556443 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with

Recording of Psychophysiological Data During Aerobatic Training

  • Upload
    erland

  • View
    213

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Recording of Psychophysiological Data During Aerobatic Training

This article was downloaded by: [Universite Laval]On: 04 October 2014, At: 07:29Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

The International Journal ofAviation PsychologyPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/hiap20

Recording ofPsychophysiological DataDuring Aerobatic TrainingNicklas Dahlstrom a , Staffan Nahlinder b , Glenn F.Wilson c & Erland Svensson ba Lund University School of Aviation , Ljungbyhed,Swedenb Swedish Defence Research Agency , Stockholm,Swedenc Air Force Research Laboratory , Wright-PattersonAir Force Base , Ohio, USAPublished online: 31 Mar 2011.

To cite this article: Nicklas Dahlstrom , Staffan Nahlinder , Glenn F. Wilson &Erland Svensson (2011) Recording of Psychophysiological Data During AerobaticTraining, The International Journal of Aviation Psychology, 21:2, 105-122, DOI:10.1080/10508414.2011.556443

To link to this article: http://dx.doi.org/10.1080/10508414.2011.556443

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness,or suitability for any purpose of the Content. Any opinions and viewsexpressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of theContent should not be relied upon and should be independently verified with

Page 2: Recording of Psychophysiological Data During Aerobatic Training

primary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly orindirectly in connection with, in relation to or arising out of the use of theContent.

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan,sub-licensing, systematic supply, or distribution in any form to anyone isexpressly forbidden. Terms & Conditions of access and use can be found athttp://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 3: Recording of Psychophysiological Data During Aerobatic Training

THE INTERNATIONAL JOURNAL OF AVIATION PSYCHOLOGY, 21(2), 105–122Copyright © 2011 Taylor & Francis Group, LLCISSN: 1050-8414 print / 1532-7108 onlineDOI: 10.1080/10508414.2011.556443

Recording of Psychophysiological DataDuring Aerobatic Training

Nicklas Dahlstrom,1 Staffan Nahlinder,2 Glenn F. Wilson,3

and Erland Svensson2

1Lund University School of Aviation, Ljungbyhed, Sweden2Swedish Defence Research Agency, Stockholm, Sweden

3Air Force Research Laboratory, Wright-Patterson Air Force Base, Ohio

Measuring pilot mental workload can be important for understanding cognitivedemands during flight involving unusual movements and attitudes. Data on heartrate, eye movements, EEG, and subjective ratings from 7 flight instructors werecollected for a flight including a repeated aerobatics sequence. Heart rate dataand subjective ratings showed that aerobatic sequences produced the highest lev-els of mental workload and that heart rate can identify low-G flight segments withhigh mental workload. Blink rate and eye movement data did not support previ-ous research regarding their relation to mental workload. EEG data were difficult toanalyze due to muscle artifacts.

Measurement of mental workload has been fundamental for understanding andimproving pilot performance and can be an important tool to reduce in-flightloss-of-control accidents, the second most important and growing category ofaviation accidents in terms of fatalities (Cox, 2005; International Civil AviationOrganization, 2004). Although mental workload has been widely used for eval-uation of aircraft design, mission analysis, and assessment of pilot performanceduring flight operations (Magnusson, 2002; Wickens & Hollands, 2000), the mostfrequently used methods have been based on subjective reporting by test pilots(Roscoe, 1992). Because psychophysiological measurements offer a nonintrusivemethod to collect continuous data about the mental workload of pilots they are

Correspondence should be sent to Nicklas Dahlstrom, Lund University School of Aviation,Drottningvagen 5, SE 260 70 Ljungbyhed, Sweden. E-mail: [email protected]

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 4: Recording of Psychophysiological Data During Aerobatic Training

106 DAHLSTROM ET AL.

of particular importance for in-flight studies (Wilson, 2002a). When an aircraft istaken to or beyond the edges of the flight envelope, psychophysiological methodscan provide useful data on pilot mental workload during the complex and dynamictask of managing direction and energy in three dimensions.

In-flight loss of control is normally the result of an airplane upset, a situationdefined as unintentional pitch attitude more than 25◦ nose up, 10◦ nose down,bank angle more than 45◦, or flight within these parameters at airspeeds inap-propriate for the conditions at hand (Carbaugh et al., 1998). Training of recoveryfrom an upset is currently not mandatory and the number of airline pilots withprevious military training, which provides experience in recovering from unusualaircraft movements and attitudes, has been steadily decreasing and is expected tocontinue to do so (Philips, 2001). Although the new Multi-Crew Pilot License (ineffect from November 2006) will ensure that future pilots receive upset recov-ery training (Hervé, 2006), the reliance on simulators for this new license andfor the whole aviation training industry might be detrimental to a pilot’s abili-ties to manage unusual aircraft movements or attitudes (Croft, 2003; Scott 2004).One way of validating simulator fidelity has been to compare psychophysiologicaldata from simulator sessions and aircraft flight, but this has been done primarilyfor military and normal civil aviation operations (Dahlstrom & Nahlinder, 2009;Magnusson, 2002). Both new and experienced airline pilots provided with upsetrecovery training in a simulator could discover that a real aircraft responds quitedifferently in an actual upset situation (Scott, 2004) because the authenticity ofsimulators in these situations might be low due to lack of data (Croft, 2003). Thishas led to introduction of new airline training programs for upset recovery (Croft,2002; Donoghue, 1995; Fiorino, 2002) and calls for mandatory aerobatics train-ing for pilots: “Training should include flights in aerobatic aircraft to practicerecovery techniques because no simulator can model the disorientation of actu-ally being upside down” (Aviation Week and Space Technology, 1995, p. 66).Psychophysiological data hold the potential to increase the knowledge of howpilots experience these situations.

Heart rate has been the psychophysiological measure most frequently usedto study the effects of flight on mental workload (Bonner & Wilson, 2002;Roscoe, 1992; Wilson, 2002c). Heart rate has proven to be a sensitive indicatorof mental workload during flight; in particular the approach and landing seg-ments have been shown to cause increased heart rate (Wilson, 2002a, 2002b).As long as the pilot is seated during flight, stable and reliable heart rate canbe obtained without interference from any significant muscular activity (Wilson,2001). However, the physical effort needed when maneuvering an aircraft closeto the edge of the flight envelope can be considerably greater than normalmaneuvering and heart rate data might then be contaminated by artifacts frommuscular activity, making them less reliable or even irrelevant for assessment

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 5: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 107

of mental workload. Heart rate has also been studied in connection with aer-obatic flight (Guézennec et al., 2001; Lang, Kessel, Denkl, & Weikl, 1976),although with a primary focus on physiological aspects rather than mentalworkload.

Other psychophysiological measures that have been used for assessment ofpilot mental workload have included heart rate variability (HRV), respiration,electrodermal activity (EDA), electroocular activity (EOG), and brain activity(EEG; Veltman, 2002; Wilson, 2001, 2002a, 2002c). Whereas HRV has beenfound less useful in the flight environment (Wilson, 2001, 2002a) EOG andEEG have the potential to be reliable measures when moderate muscular effortis involved. Neither of these two measures has been used and validated in avia-tion to the extent of heart rate, although their usefulness has been indicated byseveral studies (Wilson, 2001, 2002a, 2002c). Although there might be redun-dancy among measures, simultaneous use of a number of psychophysiologicalmeasures can facilitate determination of patterns and extraction of information(Wilson, 2001). Also, because subjective ratings can produce results contrary topsychophysiological data, the use of a number of psychophysiological measuresmight be helpful to understand such discrepancies (Wilson, 2002c).

Despite the proven usefulness of psychophysiological measurements and theawareness of the fundamental role of training within the aviation industry (Telfer,1993), it is difficult for researchers to gain access to in-flight data. Even thoughpsychophysiological methods are nonintrusive and modern recording equipmentdoes not interfere with pilot performance, aircraft are costly resources and canrarely be deployed exclusively for research purposes. This means that measure-ments must be performed in connection with flights planned for purposes otherthan research (Wilson, 2002a, 2002c) and that opportunities to study pilot mentalworkload during unusual aircraft movements or attitudes are extremely rare.

Although psychophysiological measures provide objective data of high resolu-tion in time, mean values over time periods of minutes have often been used forthe assessment of mental workload (Wilson, 2001, 2002a, 2002c). Smoothed orlower resolution has been considered useful to show changes in mental workloadbetween flight segments. However, detailed recording of specific flight segmentsthat are more limited in time or connected to individual maneuvers should be ableto produce more distinct identification of in-flight mental workload.

This research project was primarily aimed at investigating the use of psy-chophysiological methods to obtain reliable data for assessment of pilot mentalworkload during flight that includes unusual aircraft movements and attitudes(similar to those encountered during airplane upset and loss-of-control situations).Of specific interest was the potential to identify specific flight segments, in par-ticular the aerobatics sequences, and relations among heart rate, EOG, EEG, andsubjective ratings of mental workload. Also, comparison of the effect of routineflight segments known to produce high levels of mental workload (e.g., approach

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 6: Recording of Psychophysiological Data During Aerobatic Training

108 DAHLSTROM ET AL.

and landing) with the mental workload of aerobatic maneuvers should be ableto reveal the level of effects on mental workload from experiencing unusualmovements and attitudes. The results provide guidance on further developmentand use of psychophysiological methods for assessment of mental workload dur-ing unusual aircraft movements and attitudes to increase understanding of pilotmental workload in airplane upset and loss-of-control situations.

METHOD

Participants

Seven experienced male flight instructors at Lund University School of Aviationparticipated in this study (M age = 53 years, SD = 11). Two of them were aged 35and 37 and had a total of 2,170 and 2,900 flight hours respectively. The remain-ing 5 instructors were between 57 and 62 years old; 3 had more than 5,000 flighthours, 1 had 9,700 and 1 had 16,000. (Average flight time for the all of the partic-ipants was 6,793 hr, SD = 4,728.) For all of them, the greater part of their flighttime had been accumulated as flight instructors. The amount of aerobatics train-ing performed during the last 5 years varied widely among the instructors, from50 hr to 850 hr. Only 2 instructors had performed aerobatics training within thelast month (one instructor had 2.5 hr and another had 0.5 hr).

Aircraft

The flights were performed with a single-engine propeller aircraft, the ScottishAviation Bulldog, a type that had been used for training, including aerobatics, for20 years at the flight school.

Aerobatics Sequence

The aerobatics sequence selected for the experiment was used as part of the train-ing at the flight school. The sequence consisted of the following maneuvers (seeFigure 1): loop, hammerhead (stall turn), roll, wing-over, Cuban eight, Immelman(roll-off-the-top). It was performed first one time for practice and “warm up,” thentwo consecutive times with a short break with wings-level flight in between.

Psychophysiological Data

Electrocardiographic (ECG), electroencephalographic (EEG), and electrooculo-graphic (EOG) data were collected using a Vitaport II digital recording device(Vitaport Temec Instruments BV, Gemert, the Netherlands) with eight channels.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 7: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 109

Start Loop Hammerhead

Wing-over Roll

Finish

Cuban Eight Immelman

FIGURE 1 The aerobatics sequence performed in this study. The start of respectiveindividual maneuver is marked by a dot and the end by a vertical line.

The ECG data were collected by using one channel (with electrodes placed onthe sternum of the flight instructors) and run through a software program thatautomatically detects the R-peak intervals of the QRS-complex and calculates theinterbeat intervals (IBIs). One channel was used to record vertical eye movements(VEOG) and blinks with electrodes placed above and below the right eye. TheECG and EOG signals were sampled at 512 Hz, with a low-pass filter at 40 Hzand a high-pass filter at 3 Hz.

EEG data were collected using five channels and subsequently examined andcleared from artifacts. EEG signals were sampled at 256 Hz, with low-pass fil-ter at 100 Hz and high-pass filter at 3 Hz. According to the 10–20 system, thepositions Fz, Pz, P3, and Oz were used and referenced to a mastoid. A bipolarchannel with one electrode at F7 and the other approximately 3 to 4 cm anterior toC3 was also used to avoid placing an electrode under the headset band. Althoughmore channels are preferable, EEG has previously been measured for assessmentof in-flight mental workload with four channels (Hankins & Wilson, 1998) andfor mental workload and operator functional state classification with six (Wilson,Lambert, & Russell, 2000) and seven channels (Russell, Wilson, & Monett, 1996).The EEG data were sampled at 256 Hz with a band pass of 0.5 to 100 Hz, sub-mitted to spectral analysis and divided into the standard bands normally usedfor investigating it; delta (1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), and beta(13–30 Hz).

An accelerometer was used to record G-loads and these data were sampled at32 Hz. A global positioning system (GPS) receiver recorded latitude and longitudeposition as well as altitude data.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 8: Recording of Psychophysiological Data During Aerobatic Training

110 DAHLSTROM ET AL.

Questionnaires

A Likert scale ranging from 1 to 9, with anchors for the endpoints of the scale(Nahlinder, Berggren, & Persson, 2005), was used in three questionnaires to col-lect subjective ratings. Before the flight, ratings on preparedness and expectationsprior to the flight were collected. After the flight the instructors rated their mentalworkload, the success of their performance, and the experienced task difficultyduring respective flight segment, as well as for the individual maneuvers in theaerobatics sequences. The specific segments were start, transit (flight to wherethe aerobatics program was performed), sequence 1, sequence 2, return flight,approach, and landing. After the flight, another questionnaire was used for ratingson the outcome of the flight and on factors that might have affected it.

RESULTS

Preflight Questionnaire

The answers showed that the instructors were motivated and engaged for the flight,neutral on expected difficulty and expecting to be able to perform successfully.Answers to questions regarding preparations as well as personal status (physicaland mental well-being) and flight status (recent amount of flight time relevant forthe task) were neutral. The instructors believed that the success of the flight wouldbe affected by their flight status, whereas they were more neutral to the effect oftheir personal (physiological and psychological) status.

Times and G-Load for Aerobatics Sequence and Complete Flight

The aerobatics sequences were performed within 2 to 3 min (first sequenceM = 2 min, 34 sec, SD = 22 sec; second sequence M = 2 min, 30 sec, SD = 18sec). The average time for the complete flight was 28 min (SD = 9 min, 21 sec)because some flights performed the aerobatic sequences over the airport, whereassome had to fly out from the airport to find airspace to perform it. The aero-batic sequences were performed at an altitude between 2,000 and 4,000 ft and theG-load varied between 0 and 2 G.

Heart Rate Data

Heart rate data were collected and analyzed for all of the seven flight instructors.Figure 2 shows average heart rate during flight segments for the 7 instruc-tors and for the 5 instructors for whom blink rate, EOG, and EEG data wereanalyzed.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 9: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 111

80

85

90

95

100

105

110

115

120

Taxi Take-off Sequence 1 Intersequence

Sequence 2 Landing

Hea

rt r

ate

(bea

ts/m

inu

te)

Warm up

FIGURE 2 Average heart rate (beats/minute) for 5 flight instructors.

Peak heart rate averages were found for the aerobatics sequences and therewas a pronounced reduction of heart rate in between the sequences. Neither thetakeoff, warm-up, or landing segments exhibited elevated heart rates compared tothe taxi and checklist segments.

A repeated measurement analysis of variance (ANOVA) over the seven seg-ments in Figure 2 shows a significant difference over segment. Specifically the twoaerobatic sequences differ statistically from the other segments. In the aerobaticsequences, the IBIs were shorter than in the others; that is, the heart rate wasstatistically higher.

Blink Rate and VEOG

Blink rates (Figure 3) appeared to be unusually high and the raw data files werereviewed to see if the blinks in the EOG channel corresponded to EEG activ-ity. The suspect blink activity did correlate to EEG activity and thus the datareduction and analysis were completed. Data from 5 of the 7 participants wereusable. Data from the 2 remaining participants were too noisy to allow automatedor manual selection of blinks during the relevant times or to be used for EEGanalysis. Blink rates during the flight segments from 5 participants are presentedin Figure 3 (except for the warm-up segment, for which data from only 4 wereusable).

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 10: Recording of Psychophysiological Data During Aerobatic Training

112 DAHLSTROM ET AL.

0

5

10

15

20

25

30

35

40

45

50

Taxi Take-off Sequence 1 Inter

sequence

Sequence 2 Landing

Ave

rag

e b

link

rate

(b

links

/min

ute

)

Warm up

(n = 4)

FIGURE 3 Average blink rates (blinks/minute) for 5 flight instructors (4 for warm-upsegment).

VEOG Delta

13,50

14,00

14,50

15,00

15,50

16,00

16,50

Taxi Take-off Sequence 1 Inter

sequence

Sequence 2 Landing

Ave

rage

Pow

er (M

agni

tude

)

Warm up

FIGURE 4 Average vertical eye movement (VEOG) delta power magnitudes for 5 flightinstructors.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 11: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 113

The blink rates (Figure 3) were considerably higher than those found in previ-ous research (Veltman, 2002: Wilson, 2002c) for comparable segments. A paired ttest showed that only the blink rates during the taxi and takeoff segments were sta-tistically different from the others. The pattern of the VEOG data (Figure 4) wasthe same for delta, theta, alpha, and beta bands, with peak power for the warm–upand aerobatics sequences, a distinct dip in between these and low power for taxi,takeoff, and landing.

EEG

EEG data (Figure 5) showed a similar pattern in 16 out of the 20 data sets fromthe five channels in the delta, theta, alpha, and beta bands: low power for thetaxi, takeoff, and landing segments and high power for warm-up and the aerobaticsequences. All of the alpha and beta data showed this pattern and so did threeout of five channels for delta and theta (channel 3 delta showed the same patternexcept for high power for landing; channel 2 theta and delta as well as channel 3showed peak power for warm-up and declining power thereafter).

Previous research has shown that increased workload causes decreased powerin the alpha band and increased power in the delta band (Dussault, Jouanin, &Guézennec, 2004; Prinzel et al., 2003; Wilson, 2002a). The theta band has been

EEG1 Alpha

14,20

14,40

14,60

14,80

15,00

15,20

15,40

15,60

15,80

Taxi Take-off Warm up Sequence 1 Inter

sequence

Sequence 2 Landing

Ave

rag

e P

ow

er (

Mag

nit

ud

e)

FIGURE 5 Average EEG power magnitudes for the alpha band of channel 1 (electrode Fz)for 5 flight instructors.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 12: Recording of Psychophysiological Data During Aerobatic Training

114 DAHLSTROM ET AL.

implicated as increasing (Dussault et al., 2004; Hankins & Wilson, 1998) ordecreasing (Prinzel et al., 2003) as an effect of mental workload. The similari-ties in the power bands seem to indicate that the data, even after removal of eyeblinks, were contaminated by artifacts.

Subjective Ratings

All 7 flight instructors performed subjective ratings of mental workload, perfor-mance, and difficulty for seven flight segments (Figure 6) as well as for the sixindividual aerobatics maneuvers (Figure 7).

An ANOVA with flight segment as a repeated measurement showed signifi-cant effects of flight segments, F(6, 30) = 3.55, p < .01 (see Figure 6). Post-hocanalyses show that the ratings of mental workload were higher for the aerobaticssequences than for other flight segments, with landing being the highest there-after. Ratings of performance and difficulty display a similar pattern, F(6, 30) =3.27, p < .05, and F(6, 30) = 5.18, p < .001, respectively, with relatively lowerperformance ratings and higher difficulty ratings for the aerobatics sequences (seeFigure 6).

The subjective ratings of the individual maneuvers in the aerobatics sequence(Figure 7), were analyzed as a two-way repeated measurement (Sequence ×

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

Start Transit Sequence

1

Sequence

2

Return

flight

Approach Landing

Mental workload Performance Difficulty

FIGURE 6 Average ratings of mental workload, performance, and difficulty for 7 flightinstructors.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 13: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 115

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

Loop Hammerhead Roll Wing-over Cuban Eight Immelman

Sequence 1 Sequence 2

FIGURE 7 Ratings of mental workload for individual maneuvers in the aerobatic sequencesfor 7 flight instructors.

Maneuver). Sequence is the first and second time, and maneuver is the six indi-vidual maneuvers in each sequence. The results for ratings of mental workloadand difficulty both show significant main effect of maneuver: mental workload,F(5, 25) = 11.09, p < .0005; difficulty, F(5, 25) = 8.622, p < .0005, but no maineffect of sequence, nor of the interaction (Maneuver × Sequence). Post-hoc anal-yses reveal that “wing-over” has significantly lower ratings of mental workloadand difficulty than the other maneuvers (except the loop). The Cuban eight andImmelman have higher ratings of mental workload and difficulty than the othermaneuvers (except the hammerhead). For the ratings of performance, there wasno main effect of maneuver but a strong tendency for a main effect of sequence,F(1, 5) = 4.38, p < .10, and no interaction. The participants rated the performancehigher on the second sequence than on the first.

Postflight Questionnaire

The answers showed that the rating of expected difficulty and success of theflight corresponded well with the rating of experienced difficulty and success.The instructors were neutral on their personal and flight status and how it mighthave affected their performance. No time pressure was experienced during theflight. Instructors were neutral as to what extent they mentally could stay ahead of

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 14: Recording of Psychophysiological Data During Aerobatic Training

116 DAHLSTROM ET AL.

their actions and to what extent they had to perform actions step-by-step or coulduse automated action programs. The instructors were not disturbed or affectedby any other information (except for one who was called on by air traffic con-trol), the recording equipment, weather, or nausea (except for one instructor whoexperienced slight nausea).

DISCUSSION

This study has shown that psychophysiological data for assessment of mentalworkload can be collected during flight that includes unusual aircraft move-ments and attitudes. The data collection was performed safely and effectively in aday-to-day operational environment with seven flights recorded in a single day,without infringement on the performance of the flight instructors. In this envi-ronment particular attention has to be given to the electrodes, as proven by theloss of data from 2 instructors. Although access to aircraft and pilot instruc-tors is often limited for researchers, and even more so for this type of flight,this study has shown that data collection during aerobatic flight can be effec-tively performed with minimal disturbance in day-to-day operational activities ina flight training organization. This should encourage further research collabora-tion between universities and flight training organizations. Even though there werefew participants in this study (implying low reliability), the flights were performedwith real instructors in real aircraft in a real training environment (implying highvalidity).

This project has extended the use of heart rate for assessment of mental work-load in flight that includes unusual maneuvers and attitudes. Although musculareffort of normal and emergency maneuvering does not seem to have any signifi-cant effect on heart rate (Dahlstrom & Nahlinder, 2009; Fredriksson & Persson,2003), it has been unclear if heart rate would provide reliable data during advancedmaneuvers involving G-forces, such as those performed in aerobatics or loss-of-control situations. During the warm-up segment, the flight instructors performedthe same aerobatics maneuvers as during the aerobatics sequences, yet heart ratestayed at levels comparable to other nonaerobatics flight segments and far belowthe levels of heart rate displayed during aerobatics. This suggests that changes inheart rate during the aerobatic maneuvers reflect changes in mental workload andthat heart rate is also a reliable measure for low-G flight segments and maneuvers.Significant physiological effects of aerobatic flight on heart rate have been shown(Guézennec et al., 2001), but pilots were then subjected to considerably higherG-loads (–3.5–6 Gs) than the 0 to 2 Gs experienced in this study. Also, beyondphysiological factors, experience is reported to lead to an increased G-toleranceof 1 to 2 Gs (Mohler, 1972).

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 15: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 117

Whereas previous studies (Hankins & Wilson, 1998; Wilson, 2002c) haveshown distinctly increased heart rate in takeoff and landing segments as an effectof the mental workload, effects of other flight segments have not been as easilyrecognizable. In this study the increased heart rate during the aerobatics segmentsindicates that heart rate also can be useful in identifying other flight segments thatplace great cognitive demands on pilots. This increased potential for use of heartrate, the most stable and easily obtained psychophysiological measure, for assess-ment of mental workload is important for further research of pilot performanceduring unusual maneuvers and attitudes; it would facilitate collection and analysisof a sizable database of pilot performance in such situations. Also, the high levelsof heart rate seen in the aerobatic sequences imply that flight involving unusualmaneuvers and attitudes places greater demands on pilots than takeoff and landing(which has so far been considered the most demanding flight segments). Furtherinvestigation of in-flight psychophysiological responses (during unusual maneu-vers and attitudes as well as for other flight phases) might be able to identify thedemands of specific flight segments to increase understanding of mental workloaddemands on pilots.

The high overall blink rates compared to those found in previous researchcould not be explained, although further discussions with flight instructors triedto resolve this issue. Neither effects of strong sunlight, absence of helmet withsunshield or sunglasses (due to the presence of recording equipment), or effectsof G-forces were considered plausible explanations. The blinks are not believedto be artifacts, as in the VEOG they look like normal blinks and the blinkscorrelated with the EEG signal, as expected. Whereas Wilson (2002c) recordedblink rates of 10 to 25 and Veltman (2002) recorded 8 to 30 blinks per minutefor normal flight (including visual and instrument approaches), this study regis-tered blink rates of 30 to 50 blinks per minute. Blink rates normally decreaseduring visually demanding flight segments, an effect frequently seen in the land-ing segment (Wilson, 2002c). High blink rates have been explained by frequentchanges between inside and outside references, but this cannot explain the blinkrates during these flights, either for the “normal” segments (e.g., takeoff, land-ing) or for the aerobatics sequences, because outside references are mostly usedduring aerobatics and the weather-permitted visual flight. Although previousfindings have indicated that blink rates decrease as mental workload increase(Van Orden, Limbert, Makeig, & Jung, 2001), the results from this study donot support this conclusion. VEOG data showed an expected increase in ver-tical eye movement during the warm-up and the aerobatic sequences, duringwhich complex vertical motion and search for reference points (e.g., the hori-zon) are likely to cause this increase. Van Breda and Veltman (1998) foundthat individuals tend to blink during vertical eye movements. However, whereasVEOG power was almost identical for the taxi and takeoff segments for allpower bands, the blink rate for the taxi segment was high and for the takeoff

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 16: Recording of Psychophysiological Data During Aerobatic Training

118 DAHLSTROM ET AL.

low. This indicates that blink rate would not be correlated with vertical eyemovements.

The EEG data displayed similarities among the delta, theta, alpha, and betapower bands that imply that effects of mental workload are not present in thedata. The most frequently reported effect of mental workload on EEG is increasedworkload causing decreased power in the alpha band (Gevins et al., 1998; Gundel& Wilson, 1992; Hankins & Wilson, 1998; Wilson, 2001, 2002a, 2002c). Heartrate data and subjective ratings show increased workload during the aerobaticsequences, but this would then seem to be contradicted by similar increases inthe power of the alpha band. Increased workload has also been reported to causeincreased power in the theta band (Gevins et al. 1998; Hankins & Wilson, 1998)and delta band (Wilson, 2002c). The effect on the theta band has, however, beencontradicted by Prinzel et al. (2003), who concluded that, “Research has shownthat increases in arousal, attention and workload are followed by decreases in thetaand alpha power and an increase in beta power” (p. 30). This dispute has also beenreported by Brookings, Wilson, and Swain (1996). The fact that almost all of thepower bands showed similar patterns indicates that the data might have been con-taminated by artifacts. At first, head movements during the aerobatics sequenceswere suspected as the cause, but this was not verified by flight instructors. Rather,it was explained that the recommendation at the school is to keep the head asstill as possible during aerobatic maneuvers to avoid the risk of disorientation andneck injuries (advice seen also in Civil Aviation Authority, 2005). The fact thatthere is great physical strain on the head during the short and intensive period ofaerobatic maneuvers to keep it still and that peak values of the EEG power bandscoincided with the three segments (warmup and aerobatic sequences) in whichsuch maneuvers were performed implies that artifacts from neck, scalp, and facialmuscles have contaminated the data. This is supported by Jung et al. (2000), whostated that frequent muscle movements lead to an unacceptable amount of lost datadue to artifact rejection. Thornton (1996) concluded that in an individual wherethere is a high and continuous level of muscle activity the effects on EEG vari-ables will be “highly significant and dispersed throughput the measures” (p. 7).Russell, He, and Wilson (2005) described the problems of muscle artifacts (manyand widespread sources on the head, caused by both tension and movement) andconcluded that current methods for removal require offline processes and greathuman involvement, and that they reduce but do not remove the generated muscleartifacts and they also remove components of the EEG signal.

The subjective ratings support heart rate data indications that the aerobaticsequences demanded the highest levels of mental workload. The slightly highersubjective ratings do not, however, seem reflect the distinct increase in heart ratefor the aerobatic sequences. Wilson (2002c) previously found that pilots’ subjec-tive ratings of familiar maneuvers (e.g., takeoffs, landings) were low comparedto the equivalent peaks registered by heart rate data and proposed that experience

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 17: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 119

might bias the subjective rating in this way, whereas heart rate might be more sen-sitive to the demands placed on the pilots. With the unique experience of the flightinstructors in this study in mind, a similar explanation can be proposed: To themaerobatic maneuvers are part of their normal work and are systematically underes-timated compared to the actual mental effort that they have to invest in performingthem. This highlights the importance of complementing subjective methods withpsychophysiological data in investigations of mental workload.

The results of this study indicate that for aerobatic flight with low G-forcesheart rate can provide reliable psychophysiological data for assessment of men-tal workload. Artifacts due to the physiological effects of aerobatic flight seem tomake EEG less useful for this purpose. It is possible, however, that a greater num-ber of electrodes could facilitate artifact removal and provide EEG data of usefor assessment of mental workload. The heart rate data show that psychophysi-ological methods can distinctly discriminate between different flight phases andthat aerobatic flight demands mental efforts beyond the flight phases previousresearch has recognized as most demanding (i.e., takeoff and landing). Instructors’subjective ratings also discriminated between different aerobatic maneuvers andmatching these ratings with psychophysiological data of the same resolutionshould be a goal of methodological development in psychophysiology. Such a res-olution should be able to provide increased knowledge of pilot mental workloadduring both normal and abnormal flight situations.

In spite of advances in recording techniques, increased reliability and reso-lution of psychophysiological data to provide information on mental workloadduring unusual maneuvers and attitudes remains a challenge for further research.Measurements providing accurate assessments of the mental workload imposedon pilots during unusual maneuvers and attitudes should be able to support a cal-ibration of flight and simulator training to the training needs of such situations,which today are known exclusively through the experience of flight instructors.Although it might be argued that there are differences between aerobatics andupset recovery, aerobatic maneuvers performed by flight instructors provide asafe and controlled platform for data collection on how pilots experience unusualmaneuvers and attitudes. In a unique study about aerobatic flight and learn-ing Roessingh (2005) found that practicing aerobatic maneuvers on a PC-basedsimulator had no transfer to performing the maneuvers with an aircraft:

Aerobatic skills are needed for the execution of a series of complex aircraft con-trol actions when unusual attitudes and forces are being experienced. The workinghypothesis in training applications is that these skills only transfer from a situa-tion that is identical or almost identical to the in-flight situation. This is a solidhypothesis in the absence of convincing counter-evidence. Moreover, with regard tothese skills, one should not rule out the possibility of negative transfer from trainingenvironments that depart considerably from full fidelity. (p. 68)

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 18: Recording of Psychophysiological Data During Aerobatic Training

120 DAHLSTROM ET AL.

An increased number of loss-of-control accidents will have to be met witheffective training for pilots to increase their ability to avoid and manage such sit-uations. Further development of methods for obtaining relevant, high-resolutionpsychophysiological data holds the potential to provide increased knowledge onhow novice and seasoned pilots experience unusual maneuvers and attitudes. Thisknowledge can then be used to improve the training of the handling skills neededto prevent or manage loss of control situations.

REFERENCES

Aviation Week and Space Technology. (1995). Actual aerobatic training a must [Editorial]. AviationWeek and Space Technology, 142(19), 66.

Bonner, M. A., & Wilson, G. F. (2002). Heart rate measures of flight test and evaluation. TheInternational Journal of Aviation Psychology, 12(1), 63–77.

Brookings, J. B., Wilson, G. F., & Swain, C. R. (1996). Psychophysiological responses to changes inworkload during simulated air traffic control. Biological Psychology, 42(3), 361–377.

Carbaugh, D., Cashman, J., Carriker, M., Forsythe, D., Melody, T., Rockliff, L., et al.(1998). Aerodynamic principles of large airplane upsets. Aero Magazine, 3. Retrieved fromhttp://www.boeing.com/commercial/aeromagazine/aero_03/index2.html

Civil Aviation Authority. (2005). Safety sense Leaflet 19a: Aerobatics. Retrieved fromhttp://www.caa.co.uk/docs/33/SRG_GAD_WEBSSL19.PDF

Cox J. (2005, November). Upset recovery training—Why should we teach it? How do we teach?Presentation given at the European Aviation Training Symposium, Amsterdam, Netherlands.

Croft, J. W. (2002). Inflight upset training puts muscle behind the book work. Aviation Week and SpaceTechnology, 157(9), 58.

Croft, J. W. (2003). Refuse-to-crash: NASA tackles loss of control. Aerospace America, 41(3), 42–45.Dahlstrom, N., & Nahlinder, S. (2009). Mental workload in aircraft and simulator during basic civil

aviation training. International Journal of Aviation Psychology, 19, 309–325.Donoghue, J. A. (1995). Keepin’ the shiny side up. Air Transport World, 32(10), 47.Dussault, C., Jouanin, J.-C., & Guézennec, C.-Y. (2004). EEG and ECG changes during selected flight

sequences. Aviation Space and Environmental Medicine, 75(10), 889–897.Fiorino, F. (2002). Upset training aims for G-force realism. Aviation Space and Environmental

Medicine, 157(23), 60.Fredriksson, N., & Persson, M. (2003). Mätning av mental arbetsbelastning under flygning

[Measurement of mental workload during flight] [Examination paper]. Ljungbyhed, Sweden: LundUniversity School of Aviation.

Gevins, A., Smith, M. E., Leong, H., McEvoy, L., Whitfield, S., Du, R., et al. (1998). Monitoringworking memory load during computer-based tasks with EEG pattern recognition methods. HumanFactors, 40(1), 79–91.

Guézennec, C.-Y., Louisy, F., Portier, H., Laude, D., Chapuis, B., & Plésant, J. (2001). Effects of aer-obatics flight on oxygen consumption and heart rate control: Influence on autonomic cardiovascularregulation during recovery. European Journal of Applied Physiology, 84(6), 562–568.

Gundel, A., & Wilson, G. F. (1992). Topographical changes in the ongoing EEG related to the difficultyof mental tasks. Brain Topography, 5(1), 17–25.

Hankins, T. C., & Wilson, G. F. (1998). A comparison of heart rate, eye activity, EEG and subjec-tive measures of pilot mental workload during flight. Aviation Space and Environmental Medicine,69(4), 360–367.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 19: Recording of Psychophysiological Data During Aerobatic Training

RECORDING OF PSYCHOPHYSIOLOGICAL DATA 121

Hervé, X. (2006). Technological advances facilitate changes in licensing and training standards. ICAOJournal, 2(61), 22–23.

International Civil Aviation Organization. (2004). The ICAO global aviation safety plan (GASP) (2004ed.). Retrieved from http://www.icao.int/icao/en/anb/gasp/GASP.html

Jung, T. P., Makeig, S., Humphries, C., Lee, T. W., McKeown, M. J., Iragui, V., et al. (2000).Removing electroencephalographic artifacts by blind source separation. Psychophysiology, 37(2),163–178.

Lang, E., Kessel, R., Denkl, P., & Weikl, A. (1976). Changes in heart rate, blood pressure and respira-tory frequency in inexperienced persons during an aerobatic programme. Medizinishe Welt, 27(2),70–74.

Magnusson, S. (2002). Similarities and differences in psychophysiological reactions between sim-ulated and real air-to-ground missions. International Journal of Aviation Psychology, 12(1),49–61.

Mohler, S. R. (1972). G effects on the pilot during aerobatics. Retrieved from http://www.faa.gov/library/reports/medical/oamtechreports/1970s/media/AM72-28.pdf

Nahlinder, S., Berggren, P., & Persson, B. (2005, September). Increasing training efficiency usingembedded pedagogical tools in a combat flight simulator. Paper presented at the Human Factorsand Ergonomics Society annual meeting, Orlando, FL.

Philips, E. H. (2001). Upset training increases across pilot spectrum. Aviation Week and SpaceTechnology, 155(22), 72.

Prinzel, L. J., Parasuraman, R., Freeman, F. G., Scerbo, M. W., Mikulka, P. J., & Pope, A. T. (2003).Three experiments examining the use of electroencephalogram, event-related potentials, and heartrate-variability for real-time adaptive automation design (Tech. Rep. No. 2003–212 442). Hampton,VA: NASA Langley Research Center.

Roessingh, J. J. M. (2005). Transfer of manual flying skills from PC-based simulation to actual flight—Comparison of in-flight measured data and instructor ratings. International Journal of AviationPsychology, 15(1), 67–90.

Roscoe, A. H. (1992). Assessing pilot workload: Why measure heart rate, HRV and respiration?Biological Psychology, 34(2–3), 259–288.

Russell, C. A., He, P., & Wilson, G. F. (2005, July 22–27). Artifact detection and correction for opera-tor functional state estimation. Paper presented at the International Conference on Human ComputerInteraction (HCI International), Las Vegas, NV.

Russell, C. A., Wilson, G. F., & Monett, C. T. (1996, November 10–13). Mental workload classifica-tion using a back propagation neural network. Paper presented at the Artificial Neural Networks inEngineering conference, St. Louis, MO.

Scott, W. B. (2004). Rethinking upset training: Using simulators to develop upset-recovery procedurescould lead to negative training. Aviation Week and Space Technology, 160(2), 39.

Telfer, R. A. (1993). Introduction. In R. A. Telfer (Ed.), Aviation instruction and training (pp. 1–7).Aldershot, UK: Ashgate.

Thornton, K. E. (1996). On the nature of artifacting the qEEG. Journal of Neurotherapy, 1(3), 31–40.van Breda, L., & Veltman, J. A. (1998). Perspective information in a cockpit as a target acquisition

aid. Journal of Experimental Psychology-Applied, 4(1), 55–68.van Orden, K. F., Limbert, W., Makeig, S., & Jung, T. P. (2001). Eye activity correlates of workload

during a visuospatial task. Human Factors, 43(1), 111–121.Veltman, J. A. (2002). A comparative study of psychophysiological reactions during simulator and

real flight. International Journal of Aviation Psychology, 12(1), 33–48.Wickens, C. D., & Hollands, J. G. (2000). Assessing mental workload. In C. D. Wickens (Ed.),

Engineering psychology and human performance (3rd ed., pp. 459–471). Upper Saddle River, NJ:Prentice-Hall.

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014

Page 20: Recording of Psychophysiological Data During Aerobatic Training

122 DAHLSTROM ET AL.

Wilson, G. F. (2001). Psychophysiological inflight monitoring. In J. Fahrenberg & M. Myrtek (Eds.),Progress in ambulatory assessment (pp. 435–454). Seattle, WA: Hogrefe & Huber.

Wilson, G. F. (2002a). An analysis of mental workload in pilots during flight using multiplepsychophysiological measures. International Journal of Aviation Psychology, 12(1), 3–18.

Wilson, G. F. (2002b). A comparison of three cardiac ambulatory recorders using flight data.International Journal of Aviation Psychology, 12(1), 111–119.

Wilson, G. F. (2002c). Psychophysiological test methods and procedures. In S. G. Charlton & T. G.O’Brien (Eds.), Handbook of human factors testing and evaluation (2nd ed., pp. 127–156). Mahwah,NJ: Lawrence Erlbaum Associates, Inc.

Wilson, G. F., Lambert, J. D., & Russell, C. A. (2000, July 30–August 4). Performance enhancementwith real-time physiologically controlled adaptive aiding. Paper presented at the Human Factorsand Ergonomics Society annual meeting, San Diego, CA.

Manuscript first received: January 2007

Dow

nloa

ded

by [

Uni

vers

ite L

aval

] at

07:

29 0

4 O

ctob

er 2

014