Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 161 views

    Signal Detection Theory

    Uploaded by

    smklaborti
    This document discusses signal detection theory and how it can be used to measure perceptual thresholds free from individual biases. It explains how to conduct experiments that present stimuli requiring "yes" or "no" responses. Histograms can be created to represent the internal representations and responses for each stimulus. These histograms reveal the number of "yes" and "no" responses and are used to calculate metrics like hit rates, false alarm rates, and d', which is a measure of sensitivity that accounts for response bias. Signal detection theory models stimuli and responses as normally distributed and allows standardizing them to calculate d' as the difference between normalized hit and false alarm rates. It is presented as a powerful way to measure sensitivity in various testing situations while controlling

    Copyright:

    © All Rights Reserved
    Download as PPT, PDF, TXT or read online from Scribd

    Signal Detection Theory

    Uploaded by

    smklaborti
    161 views28 pages
    This document discusses signal detection theory and how it can be used to measure perceptual thresholds free from individual biases. It explains how to conduct experiments that present stimuli requiring "yes" or "no" responses. Histograms can be created to represent the internal representations and responses for each stimulus. These histograms reveal the number of "yes" and "no" responses and are used to calculate metrics like hit rates, false alarm rates, and d', which is a measure of sensitivity that accounts for response bias. Signal detection theory models stimuli and responses as normally distributed and allows standardizing them to calculate d' as the difference between normalized hit and false alarm rates. It is presented as a powerful way to measure sensitivity in various testing situations while controlling

    Original Description:

    SDT

    Copyright

    © © All Rights Reserved

    Available Formats

    PPT, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses signal detection theory and how it can be used to measure perceptual thresholds free from individual biases. It explains how to conduct experiments that present stimuli requiring "yes" or "no" responses. Histograms can be created to represent the internal representations and responses for each stimulus. These histograms reveal the number of "yes" and "no" responses and are used to calculate metrics like hit rates, false alarm rates, and d', which is a measure of sensitivity that accounts for response bias. Signal detection theory models stimuli and responses as normally distributed and allows standardizing them to calculate d' as the difference between normalized hit and false alarm rates. It is presented as a powerful way to measure sensitivity in various testing situations while controlling

    Copyright:

    © All Rights Reserved
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    161 views28 pages

    Signal Detection Theory

    Uploaded by

    smklaborti
    This document discusses signal detection theory and how it can be used to measure perceptual thresholds free from individual biases. It explains how to conduct experiments that present stimuli requiring "yes" or "no" responses. Histograms can be created to represent the internal representations and responses for each stimulus. These histograms reveal the number of "yes" and "no" responses and are used to calculate metrics like hit rates, false alarm rates, and d', which is a measure of sensitivity that accounts for response bias. Signal detection theory models stimuli and responses as normally distributed and allows standardizing them to calculate d' as the difference between normalized hit and false alarm rates. It is presented as a powerful way to measure sensitivity in various testing situations while controlling

    Copyright:

    © All Rights Reserved
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    of 28

    Signal detection theory

    Appendix

    Takashi Yamauchi
    Texas A&M University
    How do you measure the sensitivity of
    stimulus (color, sound, light) perception?

    Measure threshold
    How do you compare Gennas
    threshold and Casadys threshold?
    Do an experiment. OK, but how?

    Present many stimuli that require yes


    responses and no responses.
    Change the stimulus intensity, and calculate the
    average detection.
    Experiment

    Target = ?

    Ch 1 4
    Casa Genn
    dy a

    Fig. 1-13, p. 14
    Whats the problem with
    this?
    Genna and Casady are different.
    1. Genna likes basketball; Casady
    likes football.
    2. Genna loves surfing; Casady
    likes hunting.
    3. Genna like sushi; Casady like
    pasta.
    Compare Gennas threshold and
    Casadys threshold?

    Mentally represent the stimuli

    Yes / NO

    Casady may tend to say YES more often than


    Genna.
    Genna may tend to say NO more often than
    Casa Genn
    So, the thresholds dy a
    measured in this way
    may simply reflect how
    Genna and Casady are
    different in their
    attitudes, but not in
    their perceptual
    sensitivity per se.
    Fig. 1-13, p. 14
    The same situation arises simply by
    changing the pay-off scale of the experiment
    Rewarded when you find stimuli (e.g., car
    mechanics, psychiatrist, surgeon)

    Rewarded when you dont make mistakes


    (e.g., career advisor, drug test, judge)
    How can we measure
    Gennas and Casadys
    perceptual thresholds free
    from these variables?
    Signal Detection theory
    SDT is extremely powerful.
    It can be applied to any test situation
    that involves yes no responses.
    memory test (did you see it or not?)
    clinical test (does the drug (therapy) work
    or not?)
    Mechanical test (does this new engine work
    or not?)
    Software implementation (does this
    software give what the user wants or not).
    Signal Detection theory
    important technical terms (very important)
    SDT (conceptual
    background)
    Assume that your task is to judge whether a
    stimulus is blue or green.
    If you feel the stimulus is blue, you say yes.
    If you feel the stimulus is green, you say no.
    Further assume that you are shown two
    stimuli one at a time in 1 million trials.

    Your feeling

    Very green Very blue


    Lets record your internal representation
    of the stimuli after 1 million trials.

    Your feeling

    Very green Very blue


    Lets simulate your decision criterion.
    When you are more conservative, the bar shifts to the right.
    When you are more liberal, the bar shifts to the left.

    NO YES

    Your feeling

    Very green Very blue


    Lets create histograms for the responses
    made for the two stimuli.

    NO YES

    Your feeling

    Very green Very blue


    Lets create histograms for the responses
    made for the two stimuli.

    NO YES

    Your feeling

    Very green Very blue


    Lets create histograms for the responses
    made for the two stimuli.

    NO YES

    Your feeling

    Very green Very blue


    What do these histograms tell you?
    You have 1 million
    responses. Each bin
    represents
    the number of trials Out of 1 million
    you had a particular NO YES trials, how many
    feeling. times you felt
    very blue.

    Out of 1 million
    trials, how many
    times you felt
    very green.

    Very green Your feeling Very blue


    What do these histograms tell you?
    You have 1 million
    responses. Each bin
    represents Out of 1 million
    the number of trials trials, how many
    you had a particular NO YES times you said
    feeling. yes?

    Out of 1 million
    trials, how many
    times you said
    No?

    Very green Your feeling Very blue


    Lets generalize a bit
    how many times
    you said yes?
    You have infinitely
    many trials and your
    histograms are Given the stimuli
    divided into very were blue, how
    small bins. NO YES many times you
    said yes?

    how many times


    Given the stimuli
    you said No?
    were green, how
    many times you
    said yes?

    Your feeling

    Very green Very blue


    Stimuli were present
    (signal)

    Stimuli were absent


    (noise)

    NO YES
    Your feeling

    Very uncertain Very certain


    Calculate
    hit =
    False alarm =

    NO YES
    Your feeling

    Very uncertain Very certain


    Your hit, FA, miss, and correct rejection are still
    influenced by the decision criterion you have.
    How do you measure the sensitivity?
    How do you measure d?
    d=Z2-Z1
    Z2

    Z1

    NO YES
    Assume that both signal and
    noise are normally distributed
    (bell curves)

    Standardize the normal


    distributions N(0, 1)

    The area below the standard


    normal distribution corresponds
    to the probability.
    d = Z2 Z1

    Calculate Z1 and Z2 from hit


    and false alarm scores
    Signal Detection theory
    important technical terms (very important)
    Homework

    Calculate:
    Hit =
    False alarm=
    Miss =
    Correct rejection =
    d =

    You might also like