Presented by: Deepika Dasari Neuroergonomic Analysis of Air

Transcription

Presented by: Deepika Dasari Neuroergonomic Analysis of Air
Neuroergonomic Analysis of Air Traffic Control Performance
Part III: Assessment of mental state by EEG in ATC task
Guofa Shou1, Deepika Dasari1, Lei Ding1,2
1School of electrical and computer engineering, University of Oklahoma.
2Center of biomedical engineering, University of Oklahoma.
Presented by: Deepika Dasari
Oklahoma/Kansas Judgment and Decision Making Group Workshop- Oklahoma City, OK, May 4, 2013
Outline
• Introduction
• EEG component based index
• Low fidelity task
• EEG component identification
• Sensitivity to time on task
• Sensitivity to mental workload
• High fidelity tasks
• Summary
• Acknowledgements
Introduction
 Prolonged durations on monotonous task can lead to
degraded task performance, this phenomena is referred to
time-on-task effect [1]
 Time on task effect is related to an excessive feeling of
tiredness and reduced alertness, which impairs both capability
and willingness to perform a task [2,3]
 Workload variations also affect the task performance
 Brain activity is believed to be a sensitive measure of mental
states (i.e., workload, engagement, effort, etc.) [3]
[1] Åkerstedt, T., Kecklund, G., & Knuttson, A. (1991). Manifest sleepiness and the EEG spectral content during night work. Sleep, 14, 221–225.
[2] Johns, M. W. (2000). A sleep physiologist’s view of the drowsy driver.Transportation Research Part F, 3, 241–249. doi: org/10.1016/S1369-8478(01)00008-0
[3] Craig, A., Tran, Y., Wijesuriya, N., & Boord, P. (2006). A controlled investigation into the psychological determinants of fatigue. Biological Psychology, 72, 78–87.
doi: org/10.1016/j.biopsycho.2005.07.005
ATC tasks
 Air traffic controller (ATC) studies
 Low Fidelity (C-Team)
 High Fidelity (CRA & ATCARS)
Motor
R
R
Pre-Motor
L
L
R
Occipital
◊ Theta (4-7Hz)
(4-7Hz),Alpha (8-12Hz),Beta (12-30 Hz)
◊ Theta/Alpha, Theta/Beta
Theta/Beta, Beta/(Theta+Alpha)
L
Secondary
Somotasensory
• Different EEG indexes are
calculated in terms frequency
bands as a function of time
Frontal
• IC patterns can be divided into
different groups based on spatialspectral patterns [4]
Central
Medial
EEG component based Index
Bilateral
Tangential
Medial
[4] Shou, G., Ding, L., & Dasari, D. (2012). Probing neural activations from continuous EEG in a real-world task: Timefrequency independent component analysis. Journal of Neuroscience Methods.
Left
Right
Low fidelity: C-Team task
 C-Team (i.e., Controller Teamwork
Evaluation and Assessment
Methodology) was developed for
the training and practice of air
traffic controllers [5]
 Participants: 10
 Sessions:
• one training session (0.5 hr)
• two recording sessions (2 hrs)
 Performance data
• number of clicks
CTEAM Task
ATC Selection Laboratory
[5] Bailey, L.L., Broach, D.M., Thompson, R.C., Enos, R.J., 1999. Controller Teamwork Evaluation and Assessment Methodology: A Scenario Calibration Study.
Federal Aviation Administration Office of Aviation Medicine, Washington, DC.
EEG ICs of interest
 Group level ICA, is performed
after combining EEG data
from all sessions based on
brain’s microstates [1]
 Two ICs with evident spatial
and spectral pattern were
investigated for workload and
time-on-task effect.
[1] Lehmann D (1990) Brain electric microstates and cognition: the atoms of thought. In: E.R. John (ed): Machinery of the Mind. Birkhäuser, Boston. pp. 209-224.
Sensitivity to time on task
 The Beta/(Theta+Alpha) index is sensitive to engagement[7]
 Linear regression is performed on spectral power data from
engagement index to investigate time-on-task effect
 12 (out of 20) sessions were identified with significant negative slope
[7] Freeman F.G., Mikulka P.J., Prinzel L.J. & Scerbo M.W., “Evaluation of an adaptive automation system using three EEG indices with a visual tracking task,” in
Biological Psychology, 50, pp. 61–76, 1999
Sensitivity to workload
Correlation
 Correlation analysis between spectral power from different EEG Indexes
to number of operations (i.e. clicks) was performed to investigate
sensitivity to workload.
 Mental workload was sensitive to Theta/Alpha EEG Index, showing
significant number of detection in binomial test
Theta
No of positive
correlations
11 (out of 20 )
Theta/Alpha
Theta/Beta
15 (out of 20 )
11 (out of 20 )
High fidelity: CRA task
CRA : Conflict Resolution Advisory
 Time-on- task: a negative slope for linear  Mental workload: two EEG indexes
regression for engagement for two
Theta/Alpha, and Theta/Beta were
occipital related IC.
sensitive to workload.
Workload
Engagement Index
Correlation P
Coefficient
Theta
0.06
0.69
Theta/Alpha
0.34
1.94E-02
Theta/Beta
0.399
5.9E-03
High fidelity: ATCARS task
Engagement Index
 ATCAR: Air Traffic Control
Advanced Research Simulator
 Similar IC patterns were
identified in the high fidelity
ATCAR task
 Engagement index is sensitive
to the time-on-task effect
 Further data collection and
analysis are in progress
Participant 1
Participant 2
Participant 3
Summary
 ICA analysis was used to decompose EEG data into different
brain related ICs
 Similar spatial patterns were identified across task levels
 Component based EEG Indexes were sensitive to time-on-task
and mental workload in low-fidelity and high-fidelity tasks
 Further investigations to assess the specificity of these EEG
indices to mental effort, mental workload and mental fatigue
will be conducted
 These results can augment in developing real-time systems to
monitor mental state of ATCs
Acknowledgements
 Research Support: NSF CAREER ECCS-0955260, DOT-FAA 10-G-008,
and OCAST HR09-125S.
 Thanks to Dr. Bailey, Dr. Millan and their team at FAA-CAMI for
data collection efforts.
Contact Information:
Computational Neuroimaging & Neuroengineering
Lab (CNNL)
Address:
University of Oklahoma
3100 Monitor Ave. Suite 280
Phone: (405) 325-3774
Website: http://www.ou.edu/ouneuro/
E-mail: leiding@ou.edu, gshou@ou.edu,
deepika@ou.edu