Published on March 5, 2008
Slide1: Visualizing Uncertainty in High Time-Stress Christopher D. Wickens Head, Human Factors Division University of Illinois, Institute of Aviation Terms Model of Influences Challenge and Results of Empirical Research Human Factors Guidelines for Best Display Practices Visualizing Uncertainty in High Time-Stress: Visualizing Uncertainty in High Time-Stress Uncertainty: Spatial-temporal resolution (imprecision) categorical uncertainty Expected Value (Risk) PXV People relatively poor at utilizing probablistic information. Value dominates risk decisions. Visualizing: Displays. Multi-media (visual, auditory sounds). Not linguistic. Time Stress: Minimizing cognitive load: best outcome Example: the pilot conflict avoidance maneuver. X O1 P1 O2 P2 Slide3: Source of Uncertainty Turbulence, Winds, Future Pilot Control Actions Slide4: Attention Domain Source Display Cognition Action Choice Appropriate (Optimal) A Simple Model Slide5: Attention Deployment Domain Source Display PERCEPTION Planning COGNITION Awareness DECISION Diagnosing The Present (Imperfect Sensors) Conservative Risky State of the World Format Rendering (Modality, Spatial, Numerical) Levels (Resolution) Yes Accuracy Choice O11 O12 O21 O22 Present Unc? No Expected Case Worst Case HUMAN Predicting the future (probablistic world) The future may change C1 C2 The choice may be wrong S1 S2 Empirical Research on “What Works”: The Challenge: Empirical Research on “What Works”: The Challenge Make credible the actuarial experience of probabilities. If low probability events are part of the display rendering, they must be experienced by the participant. Rendering of p=.01 event, participant must (a) experience the event, (b) experience it 1 out of 100, or (better) 2 out of 200. Few studies exist that have: * compared uncertainty representation vs. none. * compared different formats of uncertainty representation * collected objective performance data with actuarial experience The Empirical Results: The Empirical Results Display: Uncertainty vs. “expected case” or “worst case” No effect? Wickens Gempler & Morphew. Probablistic display of predicted flight path error does not help conflict avoidance. Yeh, Merlo & Wickens. Uncertain intelligence template vs. does not improve attention allocation in military target cueing when explicitly displays degraded spatial resolution (increased position uncertainty) of the cue. Smith & Wickens: Highlighting best case, expected case, worst case outcomes does not alter NMD strategic missile launch decisions Empirical Results: Best Display Practices: Empirical Results: Best Display Practices 1. Levels of resolution: (Danger, Uncertain, Safe) (2) D-S (3) D-U-C (5) D- DU U US S More (than 2) levels help. (St. Johns and Mannes, Schinzer et al). Philosophy of “likelihood alarm”. (Sorkin & Woods). Aviation collision warnings. Why? More of the errors in a higher resolution system are not as “bad”. (This fosters greater trust in the system): How many levels needed? (Schinzer). > 4 may be all. How to render? D S D S D U S D U S Slide9: Nichols et al. Slide10: Iso-probability Plot. -40 -20 20 40 0 Iso- TTC Plot Predictive Probabilistic and Temporal Conflict Avoidance Displays (courtesy of Jason Telner & Paul Milgram, University of Toronto) (TTC = Time-to-Conflict) TTC information plotted as contour or density graph (Masked) (Masked) Empirical Format Comparisons: Empirical Format Comparisons Graphical vs. Numerical/verbal. Graphical wins (Stone et al, Kirschenbaum & Aruda, Andre & Cutler, Kirlik & Nunes) but not always (Schinzer et al) Visual vs. Auditory, Tactile: Visual wins (Basapur) Visual Spatial vs. Visual color: Spatial wins (Andre). But not always (Schinzer et al). Slide12: Stone et al. Kirchenbaum & Aruda Slide13: Schinzer et al: Investment Decisions Best Practices in Time Stressed Environments: Best Practices in Time Stressed Environments Cognitive limitations: (Sweller: Cognitive load theory) Limited time, limited expertise Extensive research on graphical presentation (Tufte, Gillen et al., Wickens & Hollands) Information overload: people will filter: what will they process? Ignore? What will decision be based on? Expected case? Worst case? What should decisions be based on? Best Practices Under Time Stress: Best Practices Under Time Stress A B vs. A B A .95 B .70 .95 .70 3M Early 3M Late 0 .05 M .95 2. Visually link uncertainty representation to uncertain element (Proximity compatibility principle): Why visual display is good. 3. Express uncertainty in the “language of action” for: DIAGNOSIS PREDICTION Spatial occupancy contours time windows 4. Need for standardization of contour level (95%?) 1. Eliminate redundant extra information (declutter) .05 M .95 Just as good? Consequences of Supporting Risk-Seeking vs. Risk Aversive Behavior: Consequences of Supporting Risk-Seeking vs. Risk Aversive Behavior What kind of behavior does displaying uncertainty induce, invite? 1. That uncertainty exists. 2. How big it is. In high time pressure designer should evaluate the worst case outcomes. Design to avoid these, presenting relatively less probablistic information as time pressure grows. The aborted takeoff decision in aviation (Inagake). Uncertain failure Takeoff (Unfliable airplane) Abort takeoff Overshoot runway end Slide17: Conclusions More research needed (Echoes calls by others) Analyze consequences of human knowing uncertainty Displaying Information will induce specific behavior in high time-stress Slide18: References Andre, A. D., & Cutler, H. A. (1998). Displaying uncertainty in advanced navigation systems. Proceedings of the Human Factors and Ergonomics Society 42nd Annual Meeting. Santa Monica, CA: Human Factors Society. Basapur, S., Bisantz, A. M., & Kesavadas, T. (2004). The effect of display modality on decision-making with uncertainty. Proceedings of the Human Factors and Ergonomics Society. Annual Meeting. Santa Monica, CA: Human Factors Society. Bisantz, A. M., Schinzing, S., & Munch, J. Uni displaying uncertainty: Investigating the effects of display format and specificity. Unpublished MS. Finger, R. & Bisantz, A. M. (2002). Utilizing graphical formats to convey uncertainty in a decision making task. Theoretical Issues in Ergonomics Science, 3(1), 1-25. Gillan, D.J., Wickens, C.D., Hollands, J.G., & Carswell, C.M. (1998). Guidelines for presenting quantitative data in HFES publications. Human Factors, 40(1), 28-41. Kirschenbaum, S. S., & Arruda, J. E. (1994). Effects of graphic and verbal probability information on command decision making. Human Factors, 36(3), 406-418. Lipkus, I. M., & Hollands, J. (2000). The visual communication of risk. Monographs of the National Cancer Institute. Bethesda, MD: National Institutes of Health. Nicholls, D., Battino, P., Marti, P., & Pozzi, S. Presenting uncertainty to controllers and pilots. Sarter, N. B., & Schroeder, B. (2001). Supporting decision-making and action selection under time pressure and uncertainty: The case of inflight icing. Human Factors, 43, 580-590. Smith, M., & Wickens, C. D. (1999). The effects of display highlighting and event history on operator decision making in a National Missile Defense system application (ARL-99-7/FED-LAB-99-4). Savoy, IL: University of Illinois, Aviation Research Laboratory. Wickens, C. D. (1996). Designing for stress. In J. Driskell & E. Salas (Eds.), Stress and human performance (pp. 279-295). Mahwah, NJ: Lawrence Erlbaum. Wickens, C. D., & Carswell, C. M. (1995). The proximity compatibility principle: Its psychological foundation and its relevance to display design. Human Factors, 37(3), 473-494. Wickens, C. D., & Dixon, S. (2005). Is there a magic number 7 (to the minus 1)? The benefits of imperfect diagnostic automation: A synthesis of the literature (AHFD-05-1/MAAD-05-1). Savoy, IL: University of Illinois, Aviation Human Factors Division. Wickens, C. D., Gempler, K., & Morphew, M. E. (2000). Workload and reliability of predictor displays in aircraft traffic avoidance. Transportation Human Factors Journal, 2(2), 99-126. Wickens, C. D., & Hollands, J. (2000). Engineering psychology and human performance (3rd ed.). Upper Saddle River, NJ: Prentice Hall. Yeh, M., Merlo J. L., Wickens, C. D. & Brandenburg, D. L. (2003). Head up versus head down: The costs of imprecision, unreliability, and visual clutter on cue effectiveness for display signaling. Human Factors, 45(3), 390-407.