King Saud University College of Engineering IE – 341: “Human Factors Engineering” Fall – 2017 (1st Sem. 1438-9H) Chapter 3. Information Input and Processing.

Slides:



Advertisements
Similar presentations
Signal Detection Theory. The classical psychophysicists believed in fixed thresholds Ideally, one would obtain a step-like change from no detection to.
Advertisements

Statistical Decision Making
PSYCHOPHYSICS What is Psychophysics? Classical Psychophysics Thresholds Signal Detection Theory Psychophysical Laws.
Sensation Perception = gathering information from the environment 2 stages: –Sensation = simple sensory experiences and translating physical energy from.
Introduction to Biomedical Statistics. Signal Detection Theory What do we actually “detect” when we say we’ve detected something?
Discrimination-Shift Problems Background This type of task has been used to compare concept learning across species as well as across a broad range of.
Psychophysics 3 Research Methods Fall 2010 Tamás Bőhm.
The Method of Constant Stimuli & Signal Detection Theory VISN2211 Sieu Khuu David Lewis.
Chapter 1 Measurement, Statistics, and Research. What is Measurement? Measurement is the process of comparing a value to a standard Measurement is the.
King Saud University College of Engineering IE – 341: “Human Factors” Fall – 2015 (1 st Sem H) Chapter 3. Information Input and Processing Part.
Signal detection theory Appendix Takashi Yamauchi Texas A&M University.
Signal Detection Theory I. Challenges in Measuring Perception II. Introduction to Signal Detection Theory III. Applications of Signal Detection Theory.
Behavioral: Basic Psychology of the Senses of the User IST 331 – Organization and Design of Information Systems: User and System Principles Instructor:
Human Capabilities Part – I. Hearing (Chapter 6*) Prepared by: Ahmed M. El-Sherbeeny, PhD 1.
Sensation Perception = gathering information from the environment 2 stages: –Sensation = simple sensory experiences and translating physical energy from.
Chapter 2: Signal Detection and Absolute Judgement
ISE Recall the HIP model. ISE Beyond sensing & perceiving …  You are sitting at lunch and hear a familiar ring tone. Is that your.
Inferential Statistics Inferential statistics allow us to infer the characteristic(s) of a population from sample data Slightly different terms and symbols.
Outline of Lecture I.Intro to Signal Detection Theory (words) II.Intro to Signal Detection Theory (pictures) III.Applications of Signal Detection Theory.
King Saud University College of Engineering IE – 341: “Human Factors” Fall – 2015 (1 st Sem H) Chapter 3. Information Input and Processing Part.
SIGNAL DETECTION THEORY  A situation is described in terms of two states of the world: a signal is present ("Signal") a signal is absent ("Noise")  You.
King Saud University College of Engineering IE – 341: “Human Factors” Spring – 2016 (2 nd Sem H) Chapter 3. Information Input and Processing Part.
King Saud University College of Engineering IE – 341: “Human Factors” Spring – 2016 (2 nd Sem H) Chapter 3. Information Input and Processing Part.
Fuzzy Signal Detection Theory: ROC Analysis of Stimulus and Response Range Effects J.L. Szalma and P.A. Hancock Department of Psychology and Institute.
Lecture 1.31 Criteria for optimal reception of radio signals.
Logic of Hypothesis Testing
Chapter 1. Introduction Prepared by: Ahmed M. El-Sherbeeny, PhD
King Saud University College of Engineering IE – 341: “Human Factors Engineering” Fall – 2016 (1st Sem H) Chapter 10. Human Control of Systems.
Signal Detection, Information Theory, and Absolute Judgment
Prepared by: Ahmed M. El-Sherbeeny, PhD
Hypothesis Testing: One Sample Cases
Chapter 3. Information Input and Processing
Discrimination learning: Introduction
King Saud University College of Engineering IE – 341: “Human Factors Engineering” Fall – 2017 (1st Sem H) Chapter 3. Information Input and Processing.
Chapter 1. Introduction Prepared by: Ahmed M. El-Sherbeeny, PhD
Chapter 1. Introduction Prepared by: Ahmed M. El-Sherbeeny, PhD
Chapter 21 More About Tests.
Chapter 3. Information Input and Processing
The Measurement of Motor Performance
King Saud University College of Engineering IE – 341: “Human Factors Engineering” Fall – 2017 (1st Sem H) Chapter 3. Information Input and Processing.
Signal Detection Theory
Origins of Signal Detection Theory
SENSATION AND PERCEPTION
Measuring Social Life: How Many? How Much? What Type?
Information Units of Measurement
Signal Detection Theory
Agenda Review homework Lecture/discussion Week 10 assignment
Unit 5: Senation & Perception Day 1: Sensory Thresholds & The Eye
Chapter 2 Human Information Processing
Introduction to Summary Statistics
Chapter 3. Information Input and Processing
Review: What influences confidence intervals?
King Saud University College of Engineering IE – 341: “Human Factors Engineering” Fall – 2017 (1st Sem H) Visual Displays of Dynamic Information.
Prepared by: Ahmed M. El-Sherbeeny, PhD
Signal Detection Theory
Chapter 4 Section 2.
Solved Problems: Signal Detection Theory
Volume 25, Issue 17, Pages R736-R739 (August 2015)
Sensation.
King Saud University College of Engineering IE – 341: “Human Factors Engineering” Fall – 2017 (1st Sem H) Visual Displays of Dynamic Information.
Chapter 12 Power Analysis.
How do we make decisions about uncertain events?
A Role for the Superior Colliculus in Decision Criteria
Physical Energy + Psychological Experience
What determines Sex Ratio in Mammals?
Signal detection theory
INSTRUCTIONS for PSYCHOMETRIC TESTS.
Normal Distributions 11-Ext Lesson Presentation Holt Algebra 2.
Chapter 3. Information Input and Processing Part – I*
Introduction To Hypothesis Testing
Presentation transcript:

King Saud University College of Engineering IE – 341: “Human Factors Engineering” Fall – 2017 (1st Sem. 1438-9H) Chapter 3. Information Input and Processing Part – 4: Signal Detection Theory Prepared by: Ahmed M. El-Sherbeeny, PhD

Contents A Model of Information Processing Perception Signal Detection Theory

A Model of Information Processing

A Model of Information Processing What are models? What are they used for? Models are “abstract representations of a system or process” Best way to evaluate models is to see what they are used for “Good” model = one that can represent behavior of actual system well Types of models: Mathematical, physical, structural, verbal Information theory: mathematical model of info. transfer in communication systems Signal Detection Theory: also mathematical model Stages of Human Information Processing: Model of human information processing (next slide) shows Major stages of human info. Processing Relationships between them Source: Sander and McCormick (Ch. 3), 1993 Abstract: i.e. theoretical

Cont. A Model of Information Processing Source: Sander and McCormick (Ch. 3), 1993 Note how model is like a black box receiving stimuli, and producing stimuli, which in turn provide feedback, which starts the cycle again Use example: when some warning light (e.g. “change oil” comes on while driving on the highway)

Perception

Perception Levels of Perception Activities Involved with Perception: Level of perception depends on: stimulus & task that person faces Forms of perception: Detection (most basic form): determine if signal is present/not Identification/Recognition: indicating in which class target belongs Multilevel classification: making abs. judgment along stimulus dimension Activities Involved with Perception: Perception involves our prior Experiences Learned associations (line connecting long-term memory – perception) Simple detection involves complex Information processing Decision making This is included within signal detection theory Source: Sander and McCormick (Ch. 3), 1993

Signal Detection Theory

Signal Detection Theory Overview of Signal Detection Theory (SDT) Involves situations where Two discrete situations exist: signal / no signal Situations cannot be easily discriminated Examples: Detecting cavity on tooth x-ray Detecting defective component in a factory Detecting rain in weather forecast Concepts Associated with SDT Concept of noise Possible outcomes Concept of response criterion Influencing the response criterion Concept of sensitivity Applications of SDT Source: Sander and McCormick (Ch. 3), 1993

Cont. Signal Detection Theory Concept of Noise Nature of Noise: Involved with any situation Interferes with detection of signal Noise is generated: Externally (e.g. false radar return on radar screen) Internally: within person (e.g. miscellaneous neural activity) Noise value: Intensity varies from low to high with time Forms normal (bell-shaped) distribution When “signal” occurs: Intensity is added to background noise Person must decide if input (what s/he senses) consists of: Only noise, or Noise + signal Source: Sander and McCormick (Ch. 3), 1993

Cont. Signal Detection Theory Possible Outcomes First note the following: Person must decide: signal occurred / did not occur (2 possibilities) There are two realities: signal did occur / did not occur Thus, there are four possible outcomes: Hit: saying “signal” where there is signal False Alarm (FA): saying “signal”/there’s no signal Miss: saying “no signal” / there is signal Correct Rejection (CR): saying “no signal” / there’s no signal Response Signal Yes No s / y s / n ns / y ns / n  Source: Sander and McCormick (Ch. 3), 1993 Response Signal Yes No HIT FA MISS CR

Cont. Signal Detection Theory Concept of Response Criterion SDT helps with understanding how detection process works Basis of SDT: People set criterion along “hypothetical continuum of sensory activity”* People then use this as a basis for making their decisions We then find out position of criterion along continuum This determines probability of four outcomes (last slide) This process is illustrated in next slide (Figure 3-3) Notes regarding Figure 3-3: Figure shows hypothetical distributions of sensory activity in cases: Only noise is present Signal is added to noise The two distributions (noise and signal + noise overlap)  noise level alone may be > signal + noise (in which case?) Source: Sander and McCormick (Ch. 3), 1993 * i.e. people (depending on their judgment/personality/data/experience, etc) decide where signal occurs along all available sensory possibilities

Cont. Signal Detection Theory Source: Sander and McCormick (Ch. 3), 1993

Cont. Signal Detection Theory Cont. Concept of Response Criterion (RC) SDT assumes person sets criterion, such that when: Level of sensory activity > set criterion  person says: “signal present” Level of sensory activity < set criterion  person says: “no signal” Probabilities of four outcomes determined based on criterion (Figure) Quantity “beta” (aka response criterion / response bias) Based on position of criterion 𝑏𝑒𝑡𝑎 = 𝑏 / 𝑎 i.e. beta is ratio of signal : noise, or ratio of height of 2 curves @ criterion Values of beta (see next slide) Beta = 1, when two distributions intersect (can you show this?) When criterion is shifted to right  Beta increases (i.e. >1.0) Person says “signal” less  hits ↓, but also FA ↓ Person is considered: “conservative” When criterion is shifted to left  Beta decreases (i.e. <1.0) Person says “signal” more  hits ↑, but also FA ↑ Person is considered: “risky” (aka “liberal”)* Source: Sander and McCormick (Ch. 3), 1993 * Q: based on the criterion shown in the figure in the last slide, is the person here considered conservative or liberal?

Cont. Signal Detection Theory Effect of changing 𝒃𝒆𝒕𝒂 criterion (constant 𝒅’) Source: Sander and McCormick (Ch. 3), 1993 What is beta in each case above? Which cases are conservative, which cases are risky (liberal)? What is optimum case? Why?

Cont. Signal Detection Theory Influencing the Response Criterion Two variables affect setting the criterion Probability of detecting a signal Costs and benefits associated with 4 possible outcomes Probability of detecting a signal: e.g. If you told dentist tooth was hurting you  Probability that you have cavity ↑  after seeing suspicious spot on x-ray  dentist will likely say: “cavity”  criterion ↓ (i.e. beta ↓)  i.e. dentist can make “risky” assessment What is cost of false alarm (saying “cavity”, when there’s no cavity)?  tooth gets drilled without need What is cost of miss (saying “no cavity”, when there is cavity)?  cavity worsens  may lose tooth So, what should dentist do? (i.e. after weighing costs?) Most likely: set low criterion  call suspicious spot: “cavity” But what if you go regularly?  he will be more conservative (why?)*  will set a high criterion, and will not call it “cavity” Source: Sander and McCormick (Ch. 3), 1993 * He will be more conservative, since he knows that even if he missed it this time, it will show up more in next visit, so he will be reluctant (i.e. more conservative) this time to drill in the fear he might drill needlessly (thus, he will balance the issue from the two sides)

Cont. Signal Detection Theory Concept of Sensitivity What is sensitivity? It is keenness / resolution of sensory system RC vs. sensitivity in SDT: they are both independent of each other How to measure sensitivity (aka: 𝑑’) Sensitivity: 𝑑’ = separation between 2 distributions (see Figure 3-3) Measured in units of standard deviation: (aka Z-score) SD’s of 2 distributions are assumed equal As separation ↑  sensitivity ↑  𝑑’ ↑ (note, best to have high 𝑑’ *) Most applications: 𝑑 ′ = 0.5−2 (see next slide) Factors affecting 𝑑 ′ Some signal systems have more noise than others, thus: As noise ↑  𝑑 ′ ↓ Also note, as signal ↓  𝑑 ′ ↓ (can you show this on Figure 3-3?) Ability of people to memorize physical characteristics of signal Memory aids  𝑑 ′ ↑ (see Figure 3-2) e.g. for dentist: 𝑑 ′ ↑ by: Using better x-ray equipment Comparing x-ray of patient with previous x-rays of cavities Source: Sander and McCormick (Ch. 3), 1993 * note, increasing 𝑑’ (i.e. less noise in general) will make it less likely to make less false alarms, less misses, more hits, more correct rejections

Cont. Signal Detection Theory Effect of changing 𝒅’ Source: Sander and McCormick (Ch. 3), 1993

Cont. Signal Detection Theory Applications of SDT Many practical applications (from various studies): Sonar target applications Industrial inspection tasks Medical Diagnosis Forensic science Eye witness testimony Air traffic control Weather forecasting Reservations SDT should not be accepted without criticism (use with “grain of salt”) Using in some situations may  invalid results Reasons: Theory developed in lab (controlled conditions/experiments) Subjects given many, many trials Controlled signals and background noise levels Some applications don’t match these conditions Source: Sander and McCormick (Ch. 3), 1993

Cont. Signal Detection Theory Source: Sander and McCormick (Ch. 3), 1993

References Human Factors in Engineering and Design. Mark S. Sanders, Ernest J. McCormick. 7th Ed. McGraw: New York, 1993. ISBN: 0-07-112826-3. Web Interface for Statistics Education (WISE). Signal Detection: Overview.* http://wise.cgu.edu/wise-tutorials/tutorial-signal-detection-theory/signal-detection-overview-2/ * Check this link for more explanation and great, interactive, solved exercises.