SlideShare a Scribd company logo
Ecological Approach to Model Human Judgment
Performance under Uncertain Environment with
Autonomous Aids
TECHLAV Annual Meeting, July 24-25, 2016
Marcia Nealy (MS student)
| Dr. Younho Seong (NCAT) | Dr. Joseph Stephens (NCAT)
Hybrid Lens Model of Operator Judgment with
Autonomous Aids
Based on the Lens model with its extension, a Hybrid Lens
model was developed to precisely identify the effect of
automated decision aids on operators’ judgment
performance, depending on the quality of decision aids. The
model identifies :
•  Human operators’ sole judgment performance
•  Decision aids’ quality of producing estimates – Decision
aids’ competence
•  Effect of decision aids’ on judgment performance
•  Correspondence between the decision aid and human
operator
Introduction
•  Simulated Aircraft Identification Task with
Automated Decision Aid
– Participants received raw information (radar, speed, altitude)
and output from ADA
• Varied Properties of the Aid
– Aid Validity (how well the aid predicts the aircraft type)
– Aid Reliability (how consistently the aid makes decisions)
– “Meta-information” – whether the aid reports on its own
performance
• Participants Completed Multiple Scenarios,
Over 2 Days
– Day 1 – No Aid | Day 2 – Aid, under different conditions
•  High(+) vs. Low (-) Validity (V), Reliability (R), Meta Information/ Understandability (U)
Technical Experimentation
Methods Results
• Brunswik (1955), Hammond, Stewart,
Brehmer, & Steinmann, (1975)
• Achievement – The Correspondence
Between Judgments, and the Environment
to be Judged, Based on Available
information (the cues)
– Example: Judge if an aircraft is hostile or friendly based on
known information about the aircraft – achievement measures
how well your judgments correspond to the true nature of the
aircraft
• Multiple Regression Based Model
Formulation Provides Measures of:
– Achievement (ra)
– Degree to which judgment policies of the person match a
normative model of the environment (G)
– Degree to which the environment is well modeled with a
linear model (Re)
– Degree to which people make consistent judgments (Rs)
• Apply This Modeling Framework to:
– Human Operator (HO) making judgments unaided, based on a
set of cues
– Automated Decision Aid (ADA) making judgments, based on a
set of cues
– Human Operator making judgments in conjunction with the
decision aid (HO-ADA), where there is an additional cue – the
output of the ADA
– Achievement of these three systems, as well as the
correspondence (or reliance/use) of the HO and the ADA, and
the difference between the unaided HO, and aided HO, can
then be determined
Research Thrust Area: 3-7
Acknowledgement
This research is supported by Air Force Research Laboratory
and OSD under agreement number FA8750-15-2-0116, and also
Army Research Laboratory through the second author.
Achievement ra
YS
Judgments
YE
Criterion
Knowledge G
Un-modeled Agreement C
Cognitive
Control,
RS
Environmental
Predictability,
RE
Cue1
Cue2
Cue3
Cue4
Δ
Residuals
β1E
β2E
β3E
β4E
ŶE
Linear Model
Output
Δ
Residuals
β1S
β2S
β3S
β4S
ŶS
Linear Model
Output
YENV
X1
YADA
X2
X3
X4
Environmental
State
Automated
Decision
Aid
CUES
Validity Understand
-ability
Reliability

More Related Content

Similar to 2. NCAT _Marcia Nealy (1)

ERAU sUAS Consumer Guide June 2016 Release
ERAU sUAS Consumer Guide June 2016 ReleaseERAU sUAS Consumer Guide June 2016 Release
ERAU sUAS Consumer Guide June 2016 Release
Jonathan Westberry, MSA
 
Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...
Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...
Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...
IJERA Editor
 
Costomization of recommendation system using collaborative filtering algorith...
Costomization of recommendation system using collaborative filtering algorith...Costomization of recommendation system using collaborative filtering algorith...
Costomization of recommendation system using collaborative filtering algorith...
eSAT Publishing House
 
A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015
A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015
A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015
Journal For Research
 
Collaborative Filtering Recommendation System
Collaborative Filtering Recommendation SystemCollaborative Filtering Recommendation System
Collaborative Filtering Recommendation System
Milind Gokhale
 
MOVIE RECOMMENDATION SYSTEM
MOVIE RECOMMENDATION SYSTEMMOVIE RECOMMENDATION SYSTEM
MOVIE RECOMMENDATION SYSTEM
IRJET Journal
 
An Adaptive Framework for Enhancing Recommendation Using Hybrid Technique
An Adaptive Framework for Enhancing Recommendation Using Hybrid TechniqueAn Adaptive Framework for Enhancing Recommendation Using Hybrid Technique
An Adaptive Framework for Enhancing Recommendation Using Hybrid Technique
ijcsit
 
Human factors, user requirements and user acceptance of Shared Automated Vehi...
Human factors, user requirements and user acceptance of Shared Automated Vehi...Human factors, user requirements and user acceptance of Shared Automated Vehi...
Human factors, user requirements and user acceptance of Shared Automated Vehi...
CREDSUK
 
Major
MajorMajor
An experimental usability_test_for_different_destination
An experimental usability_test_for_different_destinationAn experimental usability_test_for_different_destination
An experimental usability_test_for_different_destination
Uzma Abidi
 
Selection of Equipment by Using Saw and Vikor Methods
Selection of Equipment by Using Saw and Vikor Methods Selection of Equipment by Using Saw and Vikor Methods
Selection of Equipment by Using Saw and Vikor Methods
IJERA Editor
 
Prediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safetyPrediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safety
iaemedu
 
Prediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safetyPrediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safety
IAEME Publication
 
Using Feedback from Data Consumers to Capture Quality Information on Environm...
Using Feedback from Data Consumers to Capture Quality Information on Environm...Using Feedback from Data Consumers to Capture Quality Information on Environm...
Using Feedback from Data Consumers to Capture Quality Information on Environm...
Anusuriya Devaraju
 
IRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User InterestIRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User Interest
IRJET Journal
 
A Novel Latent Factor Model For Recommender System
A Novel Latent Factor Model For Recommender SystemA Novel Latent Factor Model For Recommender System
A Novel Latent Factor Model For Recommender System
Andrew Parish
 
발표 자료 - PowerPoint Presentation
발표 자료 - PowerPoint Presentation발표 자료 - PowerPoint Presentation
발표 자료 - PowerPoint Presentation
butest
 
Multi Criteria Recommender Systems - Overview
Multi Criteria Recommender Systems - OverviewMulti Criteria Recommender Systems - Overview
Multi Criteria Recommender Systems - Overview
Davide Giannico
 
A Recommender System Sensitive to Intransitive Choice and Preference Reversals
A Recommender System Sensitive to Intransitive Choice and Preference ReversalsA Recommender System Sensitive to Intransitive Choice and Preference Reversals
A Recommender System Sensitive to Intransitive Choice and Preference Reversals
csandit
 
Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...
Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...
Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...
IRJET Journal
 

Similar to 2. NCAT _Marcia Nealy (1) (20)

ERAU sUAS Consumer Guide June 2016 Release
ERAU sUAS Consumer Guide June 2016 ReleaseERAU sUAS Consumer Guide June 2016 Release
ERAU sUAS Consumer Guide June 2016 Release
 
Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...
Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...
Analyzing Specialized Views of Transportation Under Mean Safety By Using Fuzz...
 
Costomization of recommendation system using collaborative filtering algorith...
Costomization of recommendation system using collaborative filtering algorith...Costomization of recommendation system using collaborative filtering algorith...
Costomization of recommendation system using collaborative filtering algorith...
 
A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015
A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015
A REVIEW PAPER ON BFO AND PSO BASED MOVIE RECOMMENDATION SYSTEM | J4RV4I1015
 
Collaborative Filtering Recommendation System
Collaborative Filtering Recommendation SystemCollaborative Filtering Recommendation System
Collaborative Filtering Recommendation System
 
MOVIE RECOMMENDATION SYSTEM
MOVIE RECOMMENDATION SYSTEMMOVIE RECOMMENDATION SYSTEM
MOVIE RECOMMENDATION SYSTEM
 
An Adaptive Framework for Enhancing Recommendation Using Hybrid Technique
An Adaptive Framework for Enhancing Recommendation Using Hybrid TechniqueAn Adaptive Framework for Enhancing Recommendation Using Hybrid Technique
An Adaptive Framework for Enhancing Recommendation Using Hybrid Technique
 
Human factors, user requirements and user acceptance of Shared Automated Vehi...
Human factors, user requirements and user acceptance of Shared Automated Vehi...Human factors, user requirements and user acceptance of Shared Automated Vehi...
Human factors, user requirements and user acceptance of Shared Automated Vehi...
 
Major
MajorMajor
Major
 
An experimental usability_test_for_different_destination
An experimental usability_test_for_different_destinationAn experimental usability_test_for_different_destination
An experimental usability_test_for_different_destination
 
Selection of Equipment by Using Saw and Vikor Methods
Selection of Equipment by Using Saw and Vikor Methods Selection of Equipment by Using Saw and Vikor Methods
Selection of Equipment by Using Saw and Vikor Methods
 
Prediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safetyPrediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safety
 
Prediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safetyPrediction of transportation specialized views of median safety
Prediction of transportation specialized views of median safety
 
Using Feedback from Data Consumers to Capture Quality Information on Environm...
Using Feedback from Data Consumers to Capture Quality Information on Environm...Using Feedback from Data Consumers to Capture Quality Information on Environm...
Using Feedback from Data Consumers to Capture Quality Information on Environm...
 
IRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User InterestIRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User Interest
 
A Novel Latent Factor Model For Recommender System
A Novel Latent Factor Model For Recommender SystemA Novel Latent Factor Model For Recommender System
A Novel Latent Factor Model For Recommender System
 
발표 자료 - PowerPoint Presentation
발표 자료 - PowerPoint Presentation발표 자료 - PowerPoint Presentation
발표 자료 - PowerPoint Presentation
 
Multi Criteria Recommender Systems - Overview
Multi Criteria Recommender Systems - OverviewMulti Criteria Recommender Systems - Overview
Multi Criteria Recommender Systems - Overview
 
A Recommender System Sensitive to Intransitive Choice and Preference Reversals
A Recommender System Sensitive to Intransitive Choice and Preference ReversalsA Recommender System Sensitive to Intransitive Choice and Preference Reversals
A Recommender System Sensitive to Intransitive Choice and Preference Reversals
 
Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...
Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...
Selecting Best Tractor Ranking Wise by Software using MADM(Multiple –Attribut...
 

2. NCAT _Marcia Nealy (1)

  • 1. Ecological Approach to Model Human Judgment Performance under Uncertain Environment with Autonomous Aids TECHLAV Annual Meeting, July 24-25, 2016 Marcia Nealy (MS student) | Dr. Younho Seong (NCAT) | Dr. Joseph Stephens (NCAT) Hybrid Lens Model of Operator Judgment with Autonomous Aids Based on the Lens model with its extension, a Hybrid Lens model was developed to precisely identify the effect of automated decision aids on operators’ judgment performance, depending on the quality of decision aids. The model identifies : •  Human operators’ sole judgment performance •  Decision aids’ quality of producing estimates – Decision aids’ competence •  Effect of decision aids’ on judgment performance •  Correspondence between the decision aid and human operator Introduction •  Simulated Aircraft Identification Task with Automated Decision Aid – Participants received raw information (radar, speed, altitude) and output from ADA • Varied Properties of the Aid – Aid Validity (how well the aid predicts the aircraft type) – Aid Reliability (how consistently the aid makes decisions) – “Meta-information” – whether the aid reports on its own performance • Participants Completed Multiple Scenarios, Over 2 Days – Day 1 – No Aid | Day 2 – Aid, under different conditions •  High(+) vs. Low (-) Validity (V), Reliability (R), Meta Information/ Understandability (U) Technical Experimentation Methods Results • Brunswik (1955), Hammond, Stewart, Brehmer, & Steinmann, (1975) • Achievement – The Correspondence Between Judgments, and the Environment to be Judged, Based on Available information (the cues) – Example: Judge if an aircraft is hostile or friendly based on known information about the aircraft – achievement measures how well your judgments correspond to the true nature of the aircraft • Multiple Regression Based Model Formulation Provides Measures of: – Achievement (ra) – Degree to which judgment policies of the person match a normative model of the environment (G) – Degree to which the environment is well modeled with a linear model (Re) – Degree to which people make consistent judgments (Rs) • Apply This Modeling Framework to: – Human Operator (HO) making judgments unaided, based on a set of cues – Automated Decision Aid (ADA) making judgments, based on a set of cues – Human Operator making judgments in conjunction with the decision aid (HO-ADA), where there is an additional cue – the output of the ADA – Achievement of these three systems, as well as the correspondence (or reliance/use) of the HO and the ADA, and the difference between the unaided HO, and aided HO, can then be determined Research Thrust Area: 3-7 Acknowledgement This research is supported by Air Force Research Laboratory and OSD under agreement number FA8750-15-2-0116, and also Army Research Laboratory through the second author. Achievement ra YS Judgments YE Criterion Knowledge G Un-modeled Agreement C Cognitive Control, RS Environmental Predictability, RE Cue1 Cue2 Cue3 Cue4 Δ Residuals β1E β2E β3E β4E ŶE Linear Model Output Δ Residuals β1S β2S β3S β4S ŶS Linear Model Output YENV X1 YADA X2 X3 X4 Environmental State Automated Decision Aid CUES Validity Understand -ability Reliability