PSYC 530 Cognitive Engineering

1. PSYC 530Cognitive Engineering Dr. Matt Peterson

2. Syllabus • Who: Dr. Matt Peterson • Web page: www.hfac.gmu/~mpeters2 • Office hours: 11-11:50 Wednesdays • Text: Engineering Psychology and Human Performance, Wickens and Hollands • Prerequisite: Psychology lab class or consent of instructor

3. Why • To prepare incoming HFAC students by providing them with a background in cognitive theory. • To establish a basic knowledge of cognitive psychology necessary for understanding human interaction with complex systems.

4. Grading Homework 10 Paper 30 Presentation 10 Exams 1-3 50 100 points total

5. Exam weighting • Best 2 of 3 exams are weighed twice as much. • The exams are worth 50% of your final grade. • Therefore, your best two exams are worth 20% each and the remaining is worth 10% (of final grade). • Example: Exam 1 92% x 40% = 36.8% Exam 2 70% x 20% = 14.0% Exam 3 98% x 40% = 39.2% 100% 90.0% x 50 = 45 points

6. Calendar Readings I expect you to read the chapter before coming to class There will also be journal articles (later) Homework The homework is designed to give you practice and a greater understanding of the subject at hand. Therefore, it is to your benefit to do the homework. Some homework will be assigned the week before an exam. I encourage you to turn it in early so that you can receive feedback before the exam.

7. Calendar Paper & Presentation An analyses an critique of a real-world system with which humans interact. Detailed description will come later in the course. Psychonomics Conference No class on November 4th.

8. Engineering Psychology • The study of cognitive systems and how this knowledge can be applied to real-world problems. • A subset of Human Factors • Topics include: • Attention • Perception • Memory • Langauge • Decision making • Response Selection

9. Why is Engineering Psych Relevant? • Butterfly Ballots • How can ballots be designed to make it clear which hole goes with which candidate? • Obvious answers may not be ideal • Can’t assume that the user will fit the system we design…we design the system to fit the user.

10. How do we measure performance? • Measure speed • Faster = better • Measure accuracy • more accurate = better • Measure workload • How effortful does the task seem to the user? • How much does the task interfere with a concurrent task? • Less workload = better

11. How do we study Engineering Psych? • Field Studies + Naturalistic • Can be difficult • Poor experiment control • Accident/Incident Reports + Highly relevant • Small sample sizes • Poor experimental control

12. How do we study Engineering Psych? (2) • User Surveys + Provide index of user attitudes (likes/dislikes) • Adequate and unbiased samples necessary • Users don’t always prefer the best design • Task Simulations + Can be naturalistic + Allow more control than field studies - Can be difficult, expensive to implement

13. How do we study Engineering Psych? (3) • Simple Experiments + Allow good experimental control + Can be cheap • Not very realistic • Literature Reviews + Inexpensive + Tons of info available - Relevant info may be difficult to find, difficult to apply.

14. General Cognitive Model Perception Cognition Action

15. “tiger!” General Cognitive Model • Perception • From sensory stimulation to stimulus identification Stimulus Sensory transduction Identification

16. “?” General Cognitive Model • Performance limited by: • Stimulus quality (fuzzy image, dark image, only a brief glimpse) • Knowledge/familiarity (top-down) • Attention

17. General Cognitive Model • Cognition • Retrieval from memory • Comparison between display items and information in memory. • Arithmetic • Decision making • Performance limited by: • Knowledge (top-down) • Quality of input from the perceptual stage (bottom-up) • Attention (time-sharing, bottlenecks, etc.) • e.g. talking on a cell phone while driving

18. General Cognitive Model • Action • Overt responses: hand movements, eye movements, etc. • Performance limited by the ability to select and program an action (experience, attention, etc.) • Example: novice typist

19. General Performance Limits All cognitive systems can suffer from these types of limits: • Data-Limited Processing • Information input to a stage is degraded or imperfect ( garbage in, garbage out) • Example: your view of a two-lane highway on a foggy morning.

20. General Performance Limits • Data-Limited Processing • Information input to a stage is degraded or imperfect • Example: your view of a two-lane highway on a foggy morning. • Resource-Limited Processing • The system is not powerful enough to perform the operation (can be due to knowledge limits). • Example: Trying to remember a 10-digit phone number before dialing.

21. General Performance Limits • Data-Limited Processing • Information input to a stage is degraded or imperfect • Example: your view of a two-lane highway on a foggy morning. • Resource-Limited Processing • The system is not powerful enough to perform the operation. • Example: Trying to remember a 10-digit phone number before dialing. • Structural (Physiological) Limits • The system can not perform simultaneous operations. • e.g. it is harder to read a book while driving than it is to listen to a book while driving.

22. Sensation The sense organs transduce physical attributes of stimuli (luminance, pressure, saltiness) into neural signals. Somehow, the physical reality is changed into a psychological reality.

23. Psychophysics • What are the limits of sensory stimulation (detection)? • How do we determine the difference between two stimuli (discrimination)?

24. Psychophysics • Detection - what is the absolute limit of a sensory system to provide information that a stimulus is present? • Discrimination - How do we tell that two stimuli are different from each other?

25. Classical Detection and Discrimination • Absolute threshold (detection) • What is the absolute minimum that we can detect? • How bright/loud/hot does an item need to be to detect it’s presence? • Difference threshold (discrimination) • How physically different do two items need to be for a person to detect the difference? • How much heavier does an item need to be to detect the difference? • How much does it take you to notice a price increase?

26. Method of Constant Stimuli • Detection • Stimuli whose intensity brackets the (likely) threshold are presented, and subjects respond ‘yes’ or ‘no’ • Example: hearing test • Discrimination • Stimuli are presented in pairs, one of which is serves as the standard. The comparison stimulus varies. • Subjects indicate whether the comparison and standard are same or different.

27. Typical Detection Data

28. Thresholds • Notice that there is not a sharp cut-off where a stimulus is detected or not. • Since there is no sharp cut-off, we calculate the threshold as being the point where the observer responded “Yes” 50% of the time.

29. Difference Thresholds • JND - Just Noticeable Difference. The amount something needs to be changed before you notice a difference. • Weber’s Law- The size of the JND is a constant proportion of the stimulus magnitude. JND = kI k is a constant the differs depending on the sensory system I is the stimulus magnitude.

30. Difference Threshold • Weber’s Law - the amount of increase needed to detect a change is proportional to the magnitude of the initial stimulus.

31. Difference Thresholds • Example - adding $5 dollars to the price of a movie ticket is noticeable, whereas adding $5 to the price of house is not.

32. Psychological Scaling? • Example - the change in brightness from going from a 50 to a 100-watt lightbulb appears larger than when going from a 100 to a 150 watt lightbulb, despite the larger (50-watt) increase.

33. JND’s get larger as magnitude increases 50 -> 100 watts = 1000 JNDs (each JND = .05 watt) 100 -> 150 watts = 500 JNDs (each JND = .1 watt)

34. Stevens’ Law • Because it is non-linear, it is more accurate than Weber’s law. • S = kIm S = magnitude of sensation k = some constant I = physical intensity m = exponent that varies depending on modality

35. Steven’s Law(m<1)

36. Steven’s Law • The Degree of curvature depends on the modality m = 1 linear relationship (essentially Weber’s law) m < 1 increasing the magnitude has diminishing returns m > 1 small increases in magnitude lead to larger and larger perceived increases

37. Steven’s Law

38. Practical Uses • Calibrating force-feedback • Standardizing the noxiousness of pyridine emitted by pig farms. • Understanding systematic perceptual errors in graphical displays: • Compression: People underestimate area or volume, such that increase the area leads to a greater underestimation. • Expansion: People overestimate color saturation. Small increases in saturation lead to larger and larger estimates.

39. More Practical uses • Audio (log) taper potentiometers • A linear volume control would seem unnatural, and perceptually would provide an abrupt change form loud to soft. • A logarithmic (audio taper) volume control perceptually provides a more linear change in volume.

40. Take Home Messages (1) Thresholds are not all-or-none, but are soft. (2) People systematically misperceive the magnitude of a stimulus, often in a non-linear way.