modeling and measuring situation awareness in individuals and teams

27
Dynamic Decision Making Lab www.cmu.edu/ddmlab Social and Decision Sciences Department Carnegie Mellon University 1 MODELING AND MEASURING SITUATION AWARENESS IN INDIVIDUALS AND TEAMS Cleotilde Gonzalez In Collaboration with: Lelyn Saner, Octavio Juarez, Mica Endsley, Cheryl Bolstad, Haydee Cuevas, and Laura Strater

Upload: aviva

Post on 22-Feb-2016

23 views

Category:

Documents


0 download

DESCRIPTION

Modeling and Measuring Situation Awareness in Individuals and Teams. Cleotilde Gonzalez. In Collaboration with: Lelyn Saner, Octavio Juarez, Mica Endsley, Cheryl Bolstad, Haydee Cuevas , and Laura Strater. Computational Models of SA Individual aspects of SA Design aspects of SA - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Modeling and Measuring Situation Awareness in Individuals and Teams

Dynamic Decision Making Labwww.cmu.edu/ddmlab

Social and Decision Sciences DepartmentCarnegie Mellon University

1

MODELING AND MEASURING SITUATION AWARENESS IN INDIVIDUALS AND TEAMS

Cleotilde Gonzalez

In Collaboration with: Lelyn Saner, Octavio Juarez, Mica Endsley, Cheryl Bolstad, Haydee

Cuevas, and Laura Strater

Page 2: Modeling and Measuring Situation Awareness in Individuals and Teams

2

• Computational Models of SA– Individual aspects of SA– Design aspects of SA– Organizational aspects of SA

• Measures of SA– Individual SA– Shared SA

• Conclusions

Agenda

Page 3: Modeling and Measuring Situation Awareness in Individuals and Teams

Situation Awareness

• the Perception of the Elements in the Environment within a Volume of Time and Space,

• the Comprehension of their Meaning, and• the Projection of their Status in the Near Future. • Formation of SA influenced by:

Individual abilities Interactions with others Environment

Page 4: Modeling and Measuring Situation Awareness in Individuals and Teams

• Integrated theory of mind: ACT-R (Anderson & Lebiere, 1998)

– Shared attention (Juarez & Gonzalez, 2003, 2004)– Learning theory (Gonzalez, Lerch & Lebiere, 2003; Gonzalez &

Lebiere, 2005)– Representation of Recognition (Gonzalez & Quesada, 2003)– Learning and decision making in dynamic systems (Gonzalez et

al., 2003; Martin, Gonzalez & Lebiere, 2004)

• Micro and Macro Cognition: Convergence and Constraints Revealed in a Qualitative Model Comparison (Lebiere, Gonzalez & Warwick, 2009)

4

Computational Cognitive Models

Page 5: Modeling and Measuring Situation Awareness in Individuals and Teams

5

Computational Cognitive Models

Page 6: Modeling and Measuring Situation Awareness in Individuals and Teams

6

A SA meta-architecture provided a full set of cognitive models interacting with OTB, and resulting in the “commander’s SA” (Gonzalez et al., 2004; Juarez & Gonzalez, 2003)

Page 7: Modeling and Measuring Situation Awareness in Individuals and Teams

7

Computational Models of Design Aspects of SA (Juarez & Gonzalez, 2006)

Page 8: Modeling and Measuring Situation Awareness in Individuals and Teams

8

• Computational Models of SA– Individual aspects of SA– Design aspects of SA– Organizational aspects of SA

• Measures of SA– Individual SA– Shared SA

• Conclusions

Agenda

Page 9: Modeling and Measuring Situation Awareness in Individuals and Teams

Individual Measures of SA: SAGAT

• Situation Awareness Global Assessment Technique (SAGAT)• Human-in-the-loop simulation exercises• Use of SAGAT queries (from GDTA)• Stop at random times and query the user• Compare response with reality of the situation

– Examples: What is the aircraft altitude?– What is the aircraft activity in this sector (en route, inbound to airport,

outbound to airport)– Which aircraft will need a new clearance to achieve landing

requirements?• SAGAT score: accuracy of the responses

Page 10: Modeling and Measuring Situation Awareness in Individuals and Teams

Individual SA measures, learning and working memory

• Can we learn to be aware? Effects of task practice and working memory influence situation awareness (SA) - Gonzalez & Wimisberg, 2007

• How do we measure individual SA– Queries may be answered while the simulation

display is not visible or covered (Endsley, 1995) or while the display is visible, uncovered (Durso et al., 1995).

10

Page 11: Modeling and Measuring Situation Awareness in Individuals and Teams

Methods

• The design was a 2 x 18 mixed design. Participants were randomly assigned to one of two conditions (covered or uncovered display) and they were asked to run the simulation 18 times (trials).

• Individuals were asked to answer SA queries while the simulation was paused

• Participants took the Visual Span Test (VSPAN) (Shah & Miyake, 1996).

11

Page 12: Modeling and Measuring Situation Awareness in Individuals and Teams

12

Page 13: Modeling and Measuring Situation Awareness in Individuals and Teams

13

Page 14: Modeling and Measuring Situation Awareness in Individuals and Teams

Results

Page 15: Modeling and Measuring Situation Awareness in Individuals and Teams
Page 16: Modeling and Measuring Situation Awareness in Individuals and Teams
Page 17: Modeling and Measuring Situation Awareness in Individuals and Teams
Page 18: Modeling and Measuring Situation Awareness in Individuals and Teams

• The correlation between SA scores and VSPAN decreased over time

• SA scores were higher in the uncovered condition than in the covered condition– This is due mostly to perception

• The effect of practice was significant only in the covered condition, but not in the uncovered condition

18

Summary of results relevant for individual measures of SA

Page 19: Modeling and Measuring Situation Awareness in Individuals and Teams

Measures of Shared SA (Saner, Bolstad, Gonzalez & Cuevas, in press; Saner, Bolstad, Gonzalez & Cuevas, in preparation)

Ground TRUTH

X1X2X3X4X5X6X7

Person 1

X1 X3X4 X6X7X8X9

Person 2

X1X2 X4 X6 X8

Shared SA-the degree to which team members possess the same SA on shared SA requirements (i.e. on the information that they both need to know)

(Endsley,1995, 1995b; Endsley & Jones, 2001)

A good measure of shared SA needs to account for the ACCURACY

Page 20: Modeling and Measuring Situation Awareness in Individuals and Teams

Shared SA

Person 1Q1

Q2Q3Q4Q5Q6Q7

Person 2Q1

Q2Q3Q4Q5Q6Q7

SimQ1

SimQ2SimQ3SimQ4SimQ5SimQ6SimQ7

Situation Awareness Global Assessment Technique (SAGAT) - Seven queries while task is stopped - Objective knowledge of situation

Score Similarity = 1-absolute value of [(p1-p2)/(p1+p2)]Range from 0 to 1

A good measure of shared SA needs to account for the SIMILARITY

Page 21: Modeling and Measuring Situation Awareness in Individuals and Teams

Method• Training at Joint Personnel Recovery Agency (JPRA) - JFCOM• 16 servicemen, 3 DoD contractors; Age M=33.85• Randomly assigned to one of four Teams:

– Navy, Army, Special Operations, or Joint Service• Utilized Cross-Training

– Five scenarios over 3 days – Each scenario had 3 to 12 incidents– Scenarios randomly stopped 3 times for SAGAT,

Communication, and Workload measures– Received training prior to the exercise

Page 22: Modeling and Measuring Situation Awareness in Individuals and Teams

Methods and Procedure

Joint Service Cell

(p1, p2, p3, p4)

Special Operations Cell

(p13, p14, p15, p16, p17)

Army Cell

(p5, p6, p7, p8)

Navy Cell

(p9, p10, p11, p12)

• Joint Personnel Recovery Agency (JPRA) training exercise

• Four team groups (i.e. cells)• Five Predictors of Shared SA

– Experience Similarity- years in real service

– Shared JPRA Knowledge- prior experience with recovery operations

– Shared Cognitive Workload- subjective ratings, five NASA-TLX scales

– Communication Distance- inverse frequency of communication

– Organizational Hub Distance- degree of dissociation from Joint Service Cell

Page 23: Modeling and Measuring Situation Awareness in Individuals and Teams

Possible Models

Classic Hierarchy Expected

Page 24: Modeling and Measuring Situation Awareness in Individuals and Teams

Results

Regression Models of True Shared SA

True

Shared SA

F Adj.

R2

Constant Experience

Similarity

Shared

Knowledge

Workload

Similarity

Organizational

Hub Distance

Communication

Distance

OVERALL 5.11** .21 -.03 .09 .26* .08 .50** -.18

Scenario 1 2.56* .09 .02 -.07 -.02 .18 -.26* -.08

Scenario 2 1.55 .03 .39 .04 -.05 -.19 -.26* .00

Scenario 3 1.66 .05 .10 .19 .09 .04 -.26* -.02

Scenario 4 2.62* .11 .26 .08 .31* -.03 -.17 -.06

Scenario 5 5.79* .24 .42 .11 -.19 -.16 .45* -.24*

*p < .0 5

**p < .0 1

Page 25: Modeling and Measuring Situation Awareness in Individuals and Teams

Conclusions – Measures of Shared SA

• Development of a Shared SA measure must account for both, accuracy and similarity of SA between members of an organization

• As shared knowledge increased, so did shared SA. • Organizational Hub Distance (OHD) is key

predictor– Physical Distance and Joint Cell Membership

• Unexpected Role of OHD– Participants processed new information directly

Page 26: Modeling and Measuring Situation Awareness in Individuals and Teams

Possible Models

ExpectedObserved

We observed that being in branch cells was associated with higher SSA rather than being in the joint cell

Page 27: Modeling and Measuring Situation Awareness in Individuals and Teams

27

• The success of Computational Models of SA, depends on appropriate and robust measures of individual and shared SA– Although individual measures and procedures exist, there is a

huge need for defining the methods and procedures for measuring SA at the team level

• We investigated measures of SA at both, the individual and team levels– We created a shared SA measure that builds on individual SA

• Computational models of both, SA and SSA can incorporate these measures.

Conclusions