Physio behavioral Synchronicity as an Index of Teamwork (LRIR) PI: - - PowerPoint PPT Presentation

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW-2012-2994.

Physio‐behavioral Synchronicity as an Index of Teamwork

(LRIR) PI: Gregory Funke (Air Force Research Laboratory)

Senior Personnel: Adam Strang

AFOSR Program Review:

Mathematical and Computational Cognition Program Computational and Machine Intelligence Program Robust Decision Making in Human‐System Interface Program (Jan 28 – Feb 1, 2013, Washington, DC)

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW-2012-2994.

Physio‐behavioral Synchronicity (Funke)

Research Objectives:

• During cooperative behaviors,

teammates may spontaneously exhibit similarity in their physiological and behavioral responses

• Evaluate the utility of measures of

PBS as predictors of team performance and processes

DoD Benefits:

• Team formation and training
• Team monitoring
• System design & evaluation

Technical Approach:

• Monitoring of several

behavioral and physiological signals across teammates

• Application of innovative

nonlinear time series analyses Budget ($k): YR 1 YR 2 YR 3 114 109.5 109.5

Project Start Date: 01/2012 Project End Date: 01/2015 2

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW-2012-2994.

List of Project Goals 1. Assessment of the relative contributions of task factors and teamwork to measures of physio-behavioral synchronicity 2. Comparison of several statistical indices for assessing synchronicity 3. Examination of the relationship between physio- behavioral synchronicity and teamwork 4. Examination of the development and stability of synchronicity 5. Synchronicity transfer

3

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW-2012-2994.

Progress Towards Goals (or New Goals) 1. Assessment of the relative contributions…

– PBS not simply due to chance and physical task constraints – May reflect “general” team processes and coordination dynamics

2. Comparison of statistical indices…

– Nonlinear measures appear to be more sensitive

3. Relationship between PBS and teamwork…

– PBS is a moderate predictor of team performance and perceived team processes

4. Development and stability of synchronicity (Y2) 5. Synchronicity transfer (Y3)

4

1

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

What is Physio‐behavioral Synchronicity (PBS)? Statistical similarity in physiological or behavioral measures observed between two

• r more members of a team engaged in

cooperative task performance

2

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

This is Hard!

Cardiac IBIs Continuous EEG (Fz)

Do you think we should? Of course we should This is Hard!

What is Physio‐behavioral Synchronicity (PBS)?

Postural Sway

3

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

Review of Recent Literature

Shockley, Santana, & Fowler, 2003 Interpretation: Verbal Comm. Drives Postural Sway Synchronicity

Cooperative & Visual

+

Uncooperative & Visual Cooperative & No Visual

+

Uncooperative & No Visual

4

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

Henning, Boucsein, & Gil, 2001

Measured Variables

• Performance (Error/Time)
• X‐Y joystick force inputs
• Cardiac IBIs
• Electrodermal Activity (EDA)
• Respirator Rate

Statistical Analysis:

• Cross‐correlation (CC)
• Weighted Spectral Coherence

Results Decrease Time predicted by:

• Increased IBI CC/Coherence
• Increased EDA Coherence

Decreased Error predicted by:

• Increased IBI CC
• Increased EDA Coherence
• Increased Respiratory Coherence

Implication: Higher levels of PBS = Better team performance

Review of Recent Literature

5

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

Questions We Wanted to Address in Year 1:

1. What are the drivers of PBS? a) Is team PBS great than inherent (chance) PBS? b) Is team PBS primarily influence by physical task demands or interpersonal coordination? c) Is PBS unique to individual teams or does it reflect a ‘common’ interpersonal coordination? 2. Do relationships exist between PBS and team processes (cohesion, communication, etc.) and/or performance? 3. What does PBS look like in a team task with differentiated roles? 4. Are nonlinear measures more sensitive for detecting/characterizing PBS?

6

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

Next Tetrominoe Falling Tetrominoe The Well Previously Placed Tetrominoe

“Rotator” “Locator”

AFRL – Team Tetris

7

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

Methods

• 80 participants (40 men, 40 women)

– 20 same sex dyads, 40 individuals

• Completed three 20‐minute trials

– Analyzed trial 3

• Dependent measures:

– Tetris score – Cardiac IBIs (re‐sampled at 4 Hz)

• .01 Hz High‐pass filter

– Anterior‐posterior postural sway (PS; 60 Hz optical tracker)

• .03 Hz to 12 Hz Band‐pass filter

– Team Cohesion (Rozzell & Gundersen, 2003) – Trust (Naquin & Paulson, 2000) – Collective Efficacy (Riggs & Knight, 1994) – Effectiveness of Team Communication (Lyons et al., 2011)

8

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q4. Statistical Analyses

Assessed PBS in IBIs and PS

• Linear Metric

– Cross correlation (CC)

• Nonlinear Metrics

– Cross‐Fuzzy Entropy‐1 (CFEn‐1) – Cross‐Recurrence Quantification Analysis (CRQA)

• Percent Recurrence (REC)
• Percent Determinism (DET)

9

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q1a. Is team PBS greater than inherent

(chance) PBS?

Step 1: Discover the underlying noise structure of IBI and PS time series. – Apply Detrended Fluctuation Analysis (DFA) to all time series

• Mean DFA‐α IBI = 1.608
• Mean DFA‐α PS = 1.616

Step 2: Create 10,000 random pairs of generated Brown Noise Step 3: Apply synchronicity measures to Brown Noise pairs Step 4: Compare observed PBS to modeled estimates of ‘inherent’ PBS. Hypothesis: Estimates of team PBS will be greater than ‘inherent’ PBS

Surrogate Analysis for bi‐variate synchronicity measures

~ Brown Noise

10

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q1a. Is team PBS greater than inherent

(chance) PBS?

IBI Synchronicity Measure Greater than Chance?

Cross Correlation CFEn‐1 CRQA % REC % DET

PS Synchronicity Measure Greater than Chance?

Cross Correlation CFEn‐1 CRQA % REC % DET Outcomes:

• Most measures indicate that observed PBS is greater than chance
• PS CC will not be examined further in subsequent analyses

11

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q1b. Is PBS primarily influenced by physical

task demands or interpersonal coordination?

Step 1: Create ‘virtual teams’ comprised of randomly paired same‐sex individuals who performed the task (virtual‐ individual dyads) Step 2: Calculated PBS measures for virtual‐individual dyads Step 3: Compared PBS of virtual‐individual dyads to experimental teams Hypothesis: PBS of experimental teams will be greater than virtual‐individual dyads

12

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q1b. Is PBS primarily influenced by physical

task demands or interpersonal coordination?

IBI Synchronicity Measure

• Exp. Teams Greater

than Virtual‐Indiv. Teams?

Cross Correlation CFEn‐1 CRQA % REC % DET

PS Synchronicity Measure

• Exp. Teams Greater

than Virtual‐Indiv. Teams?

Cross Correlation

N/A

CFEn‐1 CRQA % REC % DET Outcomes:

• Most measures indicate that PBS in experimental teams was greater than

virtual-individual dyads

• Suggests that common physical task demands contributed little, if at all, to

estimates of team PBS

13

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q1c. Is PBS unique to individual teams or does it

reflect a ‘common’ interpersonal coordination?

Step 1: Created a 2nd set of ‘virtual teams’ (virtual‐role dyads)

• Virtual: Rot‐Loc – Pairing of data from a Rotator and Locator of

different teams

• Virtual: Rot‐Rot / Loc‐Loc – Pairing of data from two participants

in the same task role of different teams

Step 2: Compare PBS of virtual‐role dyads to experimental teams Hypotheses:

• 1. If PBS was similar between experimental and virtual‐role dyads, then

support for a ‘common’ interpersonal coordination (possibly driven by task constraints)

• 2. If PBS was higher in experimental teams, then support for unique

team coordination (driven by individual strategy or communication patterns?)

14

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

IBI Synchronicity Measure

• Exp. Teams Greater

than Virtual‐Role Dyads?

Cross Correlation

X

CFEn CRQA % REC % DET

PS Synchronicity Measure

• Exp. Teams Greater

than Virtual‐Role Dyads?

Cross Correlation

N/A

CFEn CRQA % REC % DET

• Q1c. Is PBS unique to individual teams or does it

reflect a ‘common’ interpersonal coordination?

Outcomes:

• Suggests that the crucial element was performance with a partner
• Implies that PBS was related to aspects of team processes
• Does not imply that the quality of interpersonal coordination was

indistinguishable among experimental teams

15

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q2. Relations of PBS, team processes, & performance
• Q3. PBS with differentiated roles

Tetris Score Cohes. Trust Coll. Eff. Comm. IBI CC IBI CFEn-1 IBI %REC IBI %DET PS CFEn-1 PS %REC PS %DET Tetris Score 1.00 Cohesion .26 1.00 Trust .42 .56* 1.00 Collective Eff. .28 .57* .60* 1.00 Comm. .29 .76* .78* .73* 1.00 IBI CC .36 .19

• .01

.08

• .06

1.00 IBI CFEn-1

• .47*
• .44*
• .56*
• .28
• .46*
• .12

1.00 IBI %REC

• .30
• .29
• .32
• .21
• .29
• .04

.47 1.00 IBI %DET

• .34
• .27
• .15
• .13
• .12
• .06

.39 .92* 1.00 PS CFEn-1

• .07
• .45*
• .10

.08

• .37
• .20

.08

• .13
• .16

1.00 PS %REC .12

• .45*
• .12
• .08
• .29
• .02

.14

• .01

.00 .60 1.00 PS %DET .05

• .42*
• .20

.01

• .20
• .10

.24 .08 .07 .69 .86 1.00 * p < .05

16

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

• Q3. PBS with differentiated roles
• Increases in team performance were associated

with decreases in IBI CFEn-1 synchronicity

– Supports Van Schie et al. (2008) who argued that co-actors engage in complimentary behavior (rather than mimicry) in cooperative tasks to facilitate task performance

• Also observed negative correlations between team

trust, cohesion, and communication effectiveness and IBI CFEn-1, and team cohesion and PS nonlinear measures

– Team strategies based on complimentary behavior may have performed more capably and experienced higher levels of trust, cohesion, and communication effectiveness

• Reduced PBS ≠ inherent PBS

17

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

Review of Findings

Q1. PBS driven by: – Team environment, not chance (not surprising, but important to prove) – Influenced more by interpersonal coordination dynamics than physical task demands – May reflect ‘common’ interpersonal coordination Q2. PBS is related to – Team performance – Perceptions of team processes Q3. For team tasks with differentiated roles – Direction of these relations are not unexpected (Knoblich et al., 2011) – Team PBS is still greater than inherent PBS Q4. Though not definitive, results tend to support suggestions that nonlinear CFEn‐1 may be better suited for characterizing PBS in noisy responses and complex experimental contexts

18

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

100% Power at 4 Hz 70% Power at 4 Hz 40% Power at 4 Hz 10% Power at 4 Hz

PBS Metric Comparison: A Modeling Experiment

Goal: Determine which bivariate metric characterizes coupling best when specific change is expected Part 1: Sine‐wave De‐coupling Step 1: Create 10,000 Sine Waves Step 2: Add jitter to each so that relative total power at 4 Hz is progressively reduced. Step 3: Apply bivariate coupling metrics using the original wave (template) with jitter added waves Expectation: As more jitter is added the degree

• f coupling with the original wave is

reduced Best metric is that which shows steady decoupling along with low and homogenous variance

19

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

PBS Metric Comparison: A Modeling Experiment

‐3.0 ‐2.5 ‐2.0 ‐1.5 ‐1.0 ‐0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 90 80 70 60 50 40 30 20 10 CC CFEn‐1 AMI CRQA (%REC) CRQA (%DET) % Total Power at 4 Hz for Sine Waves with Jitter Coupling Strength (Mean Z-Scores from 10,000 estimates)

General Findings

• None are

perfect

• CC is “best”
• Nonlinear

metrics are mostly equivalent

20

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

PBS Metric Comparison: A Modeling Experiment

.00 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 90 80 70 60 50 40 30 20 10 CC CFEn‐1 AMI CRQA (%REC) CRQA (%DET) % Total Power at 4 Hz for Sine Waves with Jitter Variance of Coupling Strength (SD of Z-Scores from 10,000 Coupling Estimates )

General Findings

• AMI is “best”
• CC is “worst”

21

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

Part 2: Brown Noise De‐coupling Step 1: Create 10,000 Brown Noise Signals Step 2: Add jitter to ‘whiten’ the signal Step 3: Apply bivariate coupling metrics using the original noise signals (template) with jitter added signals Expectation: As more jitter is added the degree

• f coupling with the original wave is

reduced DFA = 1. 5 DFA = 1. 4 DFA = 1. 2 DFA = 1. 3 Brown Noise Best metric is that which shows steady decoupling along with low and homogenous variance IBI time series PS time series

PBS Metric Comparison: A Modeling Experiment

22

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

‐1.0 ‐0.8 ‐0.6 ‐0.4 ‐0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.4 1.3 1.2 1.1 CC CFEn‐1 AMI CRQA (%REC) CRQA (%DET) Coupling Strength (Mean Z-Scores from 10,000 estimates) DFA of Brown Noise Time Series with Jitter

PBS Metric Comparison: A Modeling Experiment

General Findings

• All are

approximately equal

23

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.4 1.3 1.2 1.1 CC CFEn‐1 AMI CRQA (%REC) CRQA (%DET) Variance of Coupling Strength (SD of Z-Scores from 10,000 Coupling Estimates )

PBS Metric Comparison: A Modeling Experiment

General Findings

• None are

perfect

• AMI is most

steady

• CFEn actually

decreases with increased noise

• CC & %DET

worst

• AMI wins
• verall!
• Approx. DFA of Brown Noise Time Series

with Jitter

24

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

What’s Next

RESCHU

• Dyadic UAV task
• Differentiated Roles
• Navigator / Pilot
• Text Comm.
• Shared Payload Operation

PBS measures

• Cardiac IBIs
• Eye tracking / Head sway
• EEG (Fz, F3, F4, CZ, PZ)

Extension Interests

• Development
• Workload
• New Analysis Techniques
• Categorical Eye‐Track
• Categorical Comm.

25

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW‐2012‐2994.

RESCHU

Navigator Pilot

DISTRIBUTION STATEMENT A. Approved for public release by AFRL Public Affairs 05/23/12; distribution is unlimited. Document Number 88ABW-2012-2994.

List of Publications Attributed to the Grant

Peer-reviewed Publications Strang, A.J., Funke, G.J., Dukes, A.W., & Middendorf, M.S. (in revisions). Physio-behavioral synchronicity in a cooperative team task: Contributors and relations. Journal of Experimental Psychology: Human Perception and Performance. Strang, A.J., Funke, G.J., & Thomas, R.D. (accepted). Physio-behavioral synchronicity lends insight into team coordination and performance in low and high fidelity cooperative tasks: Effects of development and task

• difficulty. Proceedings of the 15th International Conference on Human-Computer Interaction.

Strang, A.J., Funke, G.J., Knott, B.A., & Warm, J.S. (2011). Physio-behavioral synchronicity as an index of processes supporting team performance. Proceedings of the Human Factors and Ergonomics Society, 55(1), 1447-1451. DOI: 10.1177/1071181311551301 Invited Lectures “Panel Discussion: Modeling the complex dynamics of teamwork.” Fifteenth International Conference on Human- Computer Interaction, Las Vegas, NV, July 21-26, 2013. “Physio-behavioral synchronicity: Drivers and relations.” Cognition, Action, & Perception Speaker Series, University of Cincinnati, Cincinnati, OH, June 1, 2012. “Physio-behavioral synchronicity as an index of team processes and performance.” Department of Psychology Colloquium, Wright-State University, Dayton, OH, October 28, 2011. Conference Poster Presentations Epling, S.L., Strang, A.J., Funke, G.J., & Warm, J.S. (accepted). Coordination temporal complexity associated with team performance in a fast-paced puzzle task. Poster session presented at the 17th International Symposium

• n Aviation Psychology, Wright-State University, Dayton, OH, May 6-9, 2013.

Epling, S.L., Kinsella, A.J., Strang, A.J., Funke, G.J., Russell, S.R., & Knott, B.A. (2011, Aug.). Physio-behavioral synchronicity associated with team performance and communication. Poster session presented at the U.S. Air- Force Research Laboratory Summer Internship Forum, Wright-Patterson AFB, OH.

6