1. Szalma & Hancock HFES Europe, 20061 Fuzzy Signal Detection Theory and Human Performance: A Review of Empirical Evidence for Model Validity J.L. Szalma and P.A. Hancock University of Central Florida

2. Szalma & Hancock HFES Europe, 20062 How Do We Assess Performance? Signal Detection Theory Hit (H)Miss (M) False Alarm (FA) Correct Rejection (CR) Response YesNo State of the World Signal Noise

3. Szalma & Hancock HFES Europe, 20063 Multiple Levels of Threat in Law Enforcement Levels of Threat MinimalLowModerateHighSevere PresenceVerbalControllingIncapacitatingDeadly Levels of Response

4. Szalma & Hancock HFES Europe, 20064 Fuzzy Signal Detection Theory Traditional Set Theory: A ∩ Ā = 0 Fuzzy Set Theory: A ∩ Ā ≠ 0 Hancock, Masalonis, & Parasuraman (2000) Parasuraman, Masalonis, & Hancock (2000)

5. Szalma & Hancock HFES Europe, 20065 Fuzzy Signal Detection Theory Events can belong to the set “signal” (s) to a degree ranging from 0 to 1 Events can belong to the set “response” (r) to a degree ranging from 0 to 1 Degrees of membership possible

6. Szalma & Hancock HFES Europe, 20066 Mapping Functions

7. Szalma & Hancock HFES Europe, 20067 Assumptions of SDT  Noise and Signal+Noise distributions are normally distributed Linear ROC  Variances of the two distributions are equal Unit slope ZHZH Z FA 0 0 1 1

8. Szalma & Hancock HFES Europe, 20068 Comparison of Fuzzy and Crisp ROC Curves CrispFuzzy

9. Szalma & Hancock HFES Europe, 20069 Comparison of Fuzzy and Crisp ROC Curves Crisp Fuzzy

10. Szalma & Hancock HFES Europe, 200610 For a task requiring temporal discrimination:  Sensitivity is higher when data analyzed with FSDT compared to traditional analysis  Fuzzy response sets yield more accurate performance assessment than binary response sets (ratings match stimulus level more closely) What do we know?

11. Szalma & Hancock HFES Europe, 200611 Application of FSDT to Tank Identification

12. Szalma & Hancock HFES Europe, 200612 Results: Fuzzy SDT Binary Response Set Data of all participants met the Gaussian and equal variance assumptions 5 Category Response Set.  The equal variance Gaussian model fit for 3 observers.  The unequal variance model fit for one observer.  Neither model fit for two observers

13. Szalma & Hancock HFES Europe, 200613 Results: Fuzzy SDT 100 Category Response Set.  Equal variance Gaussian model fit for 2 observers  Unequal variance model fit the data for one observer  Neither model fit for 3 observers

14. Szalma & Hancock HFES Europe, 200614 Comparison of Crisp and Fuzzy ROC Functions Crisp Fuzzy

15. Szalma & Hancock HFES Europe, 200615 Comparison of Crisp and Fuzzy ROC Functions Crisp Fuzzy

16. Szalma & Hancock HFES Europe, 200616 Overall Conclusions Gaussian assumption met Equal variance assumption –May be met –Depends on effect of instruction set What is a ‘fuzzy response criterion’?

17. Szalma & Hancock HFES Europe, 200617 Manipulate instruction set Manipulate stimulus distribution (‘signal rate’) Establish mapping functions using multiple methodologies Question: What is the structure of the FSDT decision space? Future Directions

18. Szalma & Hancock HFES Europe, 200618 ACKNOWLEDGEMENT This work was supported in part by the Department of Defense Multidisciplinary University Research Initiative (MURI) program administered by the Army Research Office under Grant DAAD19-01-1-0621. P.A. Hancock, Principal Investigator. The views expressed in this work are those of the authors and do not necessarily reflect official Army policy. The authors wish to thank Dr. Sherry Tove, Dr. Elmar Schmeisser, and Dr. Mike Drillings for providing administrative and technical direction for the Grant.