Chapter 2 Signal Detection and Absolute Judgement SIGNAL

Chapter 2: Signal Detection and Absolute Judgement

SIGNAL DETECTION THEORY

The Signal Detection Paradigm • Hit, misses, false alarms and correct rejections.

The Signal Detection Paradigm • Change in the evidence variable caused by a weak and strong signal. xc

The Signal Detection Paradigm • Hypothetical distribution underlying signal detection theory and sensitivity

Setting the Response Criterion: Optimally in SDT • Signal probability. Optimal beta. • Payoffs. Expected value.

Setting the Response Criterion: Optimally in SDT • Human Performance in Setting Beta. – Sluggish beta. – Relationship between obtained and optimal decision criteria.

Sensitivity d′=z(H)-z(FA)

THE ROC CURVE

Theoretical Representation • Receiver Operating Characteristic (ROC) curve

Theoretical Representation • Analysis of confidence ratings in signal detection tasks.

Theoretical Representation • Z-scores.

APPLICATIONS OF SDT

Medical Diagnosis • Disease prevalence.

Recognition Memory and Eyewitness Testimony • Relative judgment

Alarm and Alert Systems • SDT and warning signals

Alarm and Alert Systems • Alarm false alarms – Minimum safe altitude warning • Solutions: – Use multiple alarm levels – Raise automated beta slightly – Keep the human in the loop – Improve operator understanding of alarm false alarms.

VIGILANCE

Target versus non-target events • Vigilance level and vigilance decrement

Measuring Vigilance Performance • Influences on sensitivity • Changes in bias

Theories of Vigilance • Arousal theory

Theories of Vigilance • Sustained demand theory. • Expectancy theory.

Techniques to Combat the Loss of Vigilance • Increasing sensitivity: – show target examples – increase target salience – reduce the event rate – train observers • Shift in Response Criterion. – Instructions, knowledge of results, false signals, confidence levels • Other techniques – Arousal and fatigue

Application • Inside and outside the Laboratory • Examples – Situation Awareness

ABSOLUTE JUDGMENT

Information Theory • What is it? – A method for quantifying the flow of information across tasks of varying complexity – A metric for measuring the information processing efficiency of the human operator • What is information? – Reduction of uncertainty – How much more do you know about the state of the world after exposure to a piece of information, than you knew beforehand?

How Much Information? What affects the quantity of information? 1. The number of possible events that could occur, N 2. The likelihood (base probabilities) of those events, P 3. The sequential constraints (context) of the events • • Contingent probabilities Pi | X, the probability of event i given that X (the context or sequential constraint) has occurred

Quantifying Information • Bits – Total stimulus information (HS) in bits HS = log 2 N – Likely events convey less information HS = log 2 (1/Pi), where Pi is the probability of occurrence of event i – For n unequal events with unequal probabilities “weighted average” based on probability

Ideal Channel • Ideal human information channel Hs HT – No information lost HS = H T = H R • HS: stimulus information • HR: response information • HT: information transmitted by operator HR

Realistic Channels HL: information lost Noise: non-relevant information • More realistic human information channels Noise Hs HR HL HT < Hs Noise Hs HR HL HT = 0

Computing HT Noise Hs HT < Hs HR HL • For a group of events, HS and HR are determined using the equation for Have • HT is determined by: HT = HS + HR - HSR – Information dispersion: HSR represents the dispersion of stimulus-response relationships (does a particular stimulus always elicit the same responses? )

Computing HSR • Determining HSR (and hence, HT): compute average of values within matrix HSR = log 2(1/. 25) = 2 bits Engineering Psych PSY 378 S HSR = log 2(1/. 125) = 3 bits 32

Computing HT HSR = log 2(4) = 2 bits HSR = log 2(8) = 3 bits HT = HS + HR – HSR HT = 2 + 2 – 2 = 2 bits HT = 2 + 2 – 3 = 1 bit

Single Dimensions • • Experimental Results Channel capacity Bow Effect Applications

Multi-dimensional Judgment • Orthogonal/Correlated Dimensions

Multi-dimensional Judgment • Dimensional Relations: – Integral and Separable. – Garner Sort task. • Configural Dimensions – Emergent features. • Summary

Multi-dimensional Judgment

Multi-dimensional Judgment • Implications of Multi-Dimensional Absolute Judgment – Example of configural dimensions for the heights and widths of rectangles