Photo Album

Photo Album

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 1 Contents A Model of Information Processing Perception Signal Detection Theory

2 A Model of Information Processing 3 A Model of Information Processing What are models? What are they used for? o Models are abstract representations of a system or process o Best way to evaluate models is to see what they are used for o Good model = one that can represent behavior of actual system well Types of models: o Mathematical, physical, structural, verbal o Information theory: mathematical model of info. transfer in communication systems

o Signal Detection Theory: also mathematical model Stages of Human Information Processing: o Model of human information processing (next slide) shows Major stages of human info. Processing Relationships between them 4 Cont. A Model of Information Processing 5 Perception 6

Perception Levels of Perception o Level of perception depends on: stimulus & task that person faces o 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: o Perception involves our prior Experiences Learned associations (line connecting long-term memory perception) o Simple detection involves complex

Information processing Decision making o This is included within signal detection theory 7 Signal Detection Theory 8 Signal Detection Theory Overview of Signal Detection Theory (SDT) o Involves situations where Two discrete situations exist: signal / no signal Situations cannot be easily discriminated o Examples: Detecting cavity on tooth x-ray

Detecting defective component in a factory Detecting rain in weather forecast Concepts Associated with SDT 1. 2. 3. 4. 5. 6. Concept of noise Possible outcomes Concept of response criterion Influencing the response criterion Concept of sensitivity Applications of SDT

9 Cont. Signal Detection Theory 1. Concept of Noise o Nature of Noise: Involved with any situation Interferes with detection of signal o Noise is generated: Externally (e.g. false radar return on radar screen) Internally: within person (e.g. miscellaneous neural activity) o Noise value: Intensity varies from low to high with time Forms normal (bell-shaped) distribution o When signal occurs:

Intensity is added to background noise Person must decide if input (what s/he senses) consists of: o Only noise, or o Noise + signal 10 Cont. Signal Detection Theory 2. Possible Outcomes o First note the following: Person must decide: signal occurred / did not occur (2 possibilities) There are two realities: signal did occur / did not occur Respon se

Yes No Yes s/y s/n No ns / y ns / n

o Thus, there are four possible outcomes: Hit: saying signal where there is signal False Alarm (FA): saying signal/theres no signal Miss: saying no signal / there is signal Correct Rejection (CR): saying no signal / theres no signal Signal Signal Respon

se Yes No Yes HIT FA No MISS CR

11 Cont. Signal Detection Theory 3. Concept of Response Criterion o SDT helps with understanding how detection process works o 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) o Notes regarding Figure 3-3: Figure shows hypothetical distributions of sensory activity in cases: o Only noise is present

o Signal is added to noise The two distributions (noise and signal + noise overlap) o noise level alone may be > signal + noise (in which case?) 12 Cont. Signal Detection Theory 13 Cont. Signal Detection Theory 3. Cont. Concept of Response Criterion (RC) o 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)

o Quantity beta (aka response criterion / response bias) Based on position of criterion i.e. beta is ratio of signal : noise, or ratio of height of 2 curves @ criterion Values of beta (see next slide) o Beta = 1, when two distributions intersect (can you show this?) o When criterion is shifted to right Beta increases (i.e. ) Person says signal less hits , but also FA Person is considered: conservative o When criterion is shifted to left 14 Beta decreases (i.e. ) Cont. Signal Detection Theory Effect of changing

criterion (constant ) 15 Cont. Signal Detection Theory 4. Influencing the Response Criterion o Two variables affect setting the criterion 1. Probability of detecting a signal 2. Costs and benefits associated with 4 possible outcomes 1. 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 2. Costs and benefits associated with 4 possible outcomes What is cost of false alarm (saying cavity, when theres no cavity)? o tooth gets drilled without need What is cost of miss (saying no cavity, when there is cavity)? o cavity worsens may lose tooth So, what should dentist do? (i.e. after weighing costs?) o Most likely: set low criterion call suspicious spot: cavity o But what if you go regularly? he will be more conservative 16 Cont. Signal Detection Theory 5. Concept of Sensitivity o What is sensitivity? It is keenness / resolution of sensory system o RC vs. sensitivity in SDT: they are both independent of each other o How to measure sensitivity (aka: )

Sensitivity: = separation between 2 distributions (see Figure 3-3) Measured in units of standard deviation: (aka Z-score) SDs of 2 distributions are assumed equal As separation sensitivity (note, best to have high *) Most applications: (see next slide) o Factors affecting 1. Some signal systems have more noise than others, thus: o As noise o Also note, as signal (can you show this on Figure 3-3?) 2. Ability of people to memorize physical characteristics of signal o Memory aids (see Figure 3-2) o e.g. for dentist: by: Using better x-ray equipment 17 Cont. Signal Detection Theory

Effect of changing 18 Cont. Signal Detection Theory 6. Applications of SDT o Many practical applications (from various studies): Sonar target applications Industrial inspection tasks Medical Diagnosis Forensic science Eye witness testimony Air traffic control Weather forecasting

o Reservations SDT should not be accepted without criticism (use with grain of salt) o Using in some situations may invalid results Reasons: o Theory developed in lab (controlled conditions/experiments) o Subjects given many, many trials o Controlled signals and background noise levels o Some applications dont match these conditions 19 Cont. Signal Detection Theory 20 o

o 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 :// etection-theory/signal-detection-overview-2 / 21

Recently Viewed Presentations

  • Meta Ethics The Last Module eeeeek! Ethical Language

    Meta Ethics The Last Module eeeeek! Ethical Language

    Naturalism A key question when trying to reach a definition of morality etc. is the is-ought gap (or the Naturalistic Fallacy). Naturalistic theories of ethics attempt to define good in terms of something that can be identified in the world...
  • Lecture3 GenMed_2nd semester - Masarykova univerzita

    Lecture3 GenMed_2nd semester - Masarykova univerzita

    Lecture 3 ESS_2nd semester Embryology - definition, the object and significance of embryology ... during the day 5 - gastrula = the bilaminar pre-embryonic stage, in which the main body axis and polarity begin to become apparent, the bilaminar stage...
  • MDE Planning and Evaluation Tools

    MDE Planning and Evaluation Tools

    Build . Profile. Get Ready. Introduction. In the previous presentation you learned about the Gather stage of the Michigan Continuous School Improvement Process (MI-CSI) and the first component - Getting Ready. Once the Get Ready step is completed, the next...
  • The Cell Organelles - KSU

    The Cell Organelles - KSU

    Prokaryotic and eukaryotic cells differ in size and complexity All cells are surrounded by a plasma membrane غشاء بلازمى. The semi-fluid substance المادة النصف سائلة within the cell is called "cytosol", السيتوبلازم containing the cell organelles عِضيات الخلية.
  • Chemical studies of rutherfordium at JAER -

    Chemical studies of rutherfordium at JAER -

    Chemical studies of rutherfordium (Rf, Z = 104) at JAEA Experimental approach to Rf chemistry Schematic flow of the experiment AIDA (Automated Ion-exchange separation apparatus coupled with the Detection system for Alpha-spectroscopy) Fluoride complex formation Acknowledgement Radial wave functions of...
  • The Pacific States/B.C. Oil Spill Task Force  B.C.

    The Pacific States/B.C. Oil Spill Task Force B.C.

    This pilot project aims at creating a national standard for marine Area Response Planning, which includes a risk assessment component. For the Pacific Region initiative, the pilot ARP area includes the Strait of Georgia and the Strait of Juan de...
  • The Cognition Hypothesis of Task-based Language Learning:

    The Cognition Hypothesis of Task-based Language Learning:

    Re-thinking for speaking, the Cognition Hypothesis, task classification and task sequencing Peter Robinson, Aoyama Gakuin University [email protected] The Cognition Hypothesis The fundamental pedagogic claim of the Cognition Hypothesis is that pedagogic tasks should be sequenced for learners in an order...
  • Under Pressure: Work-related stress in academic libraries

    Under Pressure: Work-related stress in academic libraries

    Not everyone agree that libraries are stressful places to work. According to this editorial, written shortly after the Jobs Rated Almanac (St. Martin's Griffin) was published. This book ranked 250 careers by 6 factors: environment, income, outlook, physical demands, security,...