signal detection theory
DESCRIPTION
Signal Detection Theory. I. Challenges in Measuring Perception II. Introduction to Signal Detection Theory III. Applications of Signal Detection Theory. Part 1. Challenges in Measuring Perception. Psychophysics. Psychophysics is the science of establishing quantitative relations - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/1.jpg)
Signal Detection Theory
I. Challenges in Measuring Perception
II. Introduction to Signal Detection Theory
III. Applications of Signal Detection Theory
![Page 2: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/2.jpg)
Part 1
Challenges in Measuring Perception
![Page 3: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/3.jpg)
Psychophysics
Psychophysics is the science of establishing quantitative relations between physical stimulation and perceptual events.
![Page 4: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/4.jpg)
The Method of Limits
Experimenter adjusts intensityuntil the stimulus is detected.
![Page 5: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/5.jpg)
The Method of Limits
Advantage: Measurements are madequickly.
Disadvantage: It is not clear exactlywhat’s being measured(no control for bias).
![Page 6: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/6.jpg)
“Catch Trials”
The subject is asked to make a responsewhen no stimulus has been presented(also called “noise only” trials).
![Page 7: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/7.jpg)
Not All Errors Are Equal
1. Reporting stimulus is present when it’s absent (“false alarm”).
Versus
2. Reporting stimulus is absentwhen it’s present (“miss”).
![Page 8: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/8.jpg)
Correct Responses Differ, Too
1. Reporting stimulus is present when it’s present (“hit”).
Versus
2. Reporting stimulus is absentwhen it’s absent (“correct rejection”).
![Page 9: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/9.jpg)
Stimulus-Response Matrix
Response
Sti
mu
lus
“No” “Yes”
Pre
sen
tA
bse
nt
Miss
CorrectRejection
Hit
FalseAlarm
![Page 10: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/10.jpg)
Part II
Introduction to Signal Detection Theory
S.D.T. In Words
![Page 11: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/11.jpg)
Signal Detection Theory
S.D.T. is a procedure for measuringsensitivity to stimulation, independent of the subject’s response bias.
![Page 12: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/12.jpg)
Signal Detection Theory
S.D.T. reduces the stimulus-responsematrix to two meaningful quantities.
1. Detectability (d’) - a subject’s sensitivity to stimulation.
2. Criterion () - a subject’s inclination to favor a particular response; bias.
![Page 13: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/13.jpg)
Part II
Introduction to Signal Detection Theory
S.D.T. In Pictures
![Page 14: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/14.jpg)
Distributions of Sensory ResponsesP
roba
bili
ty
Level Of Neural Activity (Arbitrary Units)
![Page 15: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/15.jpg)
Distributions of Sensory ResponsesP
roba
bili
ty
Level Of Neural Activity (Arbitrary Units)
Spontaneous Activity is Constant
![Page 16: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/16.jpg)
Distributions of Sensory ResponsesP
roba
bili
ty
Level of Neural Activity (Arbitrary Units)
Spontaneous Activity is Normally Distributed
The “Noise”Distribution
![Page 17: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/17.jpg)
Distributions of Sensory Responses
The “Noise”Distribution
The “Signal + Noise” Distribution
A Mild Stimulus is Presented (d’=1)
Pro
babi
lity
Level of Neural Activity (Arbitrary Units)
d'
![Page 18: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/18.jpg)
Distributions of Sensory Responses
The “Noise”Distribution
The “Signal + Noise” Distribution
A Moderate Stimulus is Presented (d’=2)
Pro
babi
lity
Level of Neural Activity (Arbitrary Units)
d'
![Page 19: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/19.jpg)
Distributions of Sensory ResponsesP
roba
bili
ty
Level of Neural Activity (Arbitrary Units)
d'
The “Noise”Distribution
The “Signal + Noise” Distribution
An Intense Stimulus is Presented (d’=3)
![Page 20: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/20.jpg)
Distributions of Sensory Responses
Sub-Threshold Stimulus is Presented (d’=0)
Pro
babi
lity
Level of Neural Activity (Arbitrary Units)
The “Noise”Distribution
The “Signal + Noise” Distribution
![Page 21: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/21.jpg)
Pro
babi
lity
Level of Neural Activity (Arbitrary Units)
"No, I don'tsee it"
"Yes,I see it"
Criterion
The “Noise”Distribution
The “Signal + Noise” Distribution
![Page 22: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/22.jpg)
Neutral Criterion
The “Noise”Distribution
The “Signal + Noise” Distribution
Pr
Pr
of S
+N
Neural Activity"No" "Yes"
Hits Misses
Pr
of N False
Alarms
CorrectRejections
.5
.5
![Page 23: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/23.jpg)
Liberal (low) CriterionP
rP
r of
S+
N
Neural Activity"No" "Yes"
Hits Misses
Pr
of N False
Alarms
CorrectRejections
The “Noise”Distribution
The “Signal + Noise” Distribution .2
.6
![Page 24: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/24.jpg)
Conservative (high) Criterion
The “Noise”Distribution
The “Signal + Noise” Distribution
Pr
of S
+N
Neural Activity
Hits Misses
"No" "Yes"
Pr
Pr
of N False
Alarms
CorrectRejections
.6
.2
![Page 25: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/25.jpg)
Receiver Operating Space
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
![Page 26: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/26.jpg)
Receiver Operating Characteristics
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
d’=0
![Page 27: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/27.jpg)
R.O.C. Curves
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
d’=1
d’=0
![Page 28: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/28.jpg)
R.O.C. Curves
d’=1
d’=0
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
![Page 29: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/29.jpg)
R.O.C. Curves
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
d’=1
d’=2
d’=0
![Page 30: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/30.jpg)
R.O.C. Curves
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
d’=1
d’=2d’=
3
d’=0
![Page 31: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/31.jpg)
R.O.C. Curves
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
?
![Page 32: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/32.jpg)
R.O.C. Curves
0
1
Pro
por
tion
of
Hit
s
0 1Proportion of False Alarms
d’ = -1
d’ = -2
d’ = -3
![Page 33: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/33.jpg)
Part II
Introduction to Signal Detection Theory
S.D.T. In Numbers
![Page 34: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/34.jpg)
Normal Distributions
S.D.T. is based on normal distributions.
Each normal distribution is described bya mean and a standard deviation.
![Page 35: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/35.jpg)
Normal Distributions
A given point on a normal distributioncan be described be described 3 ways.
1. Percentile (also proportion)
2. Z-score (# of standard deviations)
3. Probability Density (likelihood)
![Page 36: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/36.jpg)
Computing Detectability
d’ = zHits - zFalse Alarms
In excel, the “normsinv” function is used: Input = proportion Output = z-Score
Conceptually, detectability (d’) increaseswith the area under the R.O.C. curve.
![Page 37: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/37.jpg)
Computing Criterion
= Hit Density / False Alarm Density
In excel, the “normsdist” function is used: Input = z-Score Output = density
Conceptually, is equal to the slopeof the R.O.C. curve at single point.
![Page 38: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/38.jpg)
Part III
Applications of Signal Detection Theory
![Page 39: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/39.jpg)
S.D.T. Applications
S.D.T. can be used in perceptualdiscrimination experiments.
![Page 40: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/40.jpg)
S.D.T. And DiscriminationP
roba
bili
ty
Perceived Speed
"No, 2nd Stimuluswas not faster"
"Yes,2nd stimuluswas faster"
The “slow”distribution
The “fast”distribution
![Page 41: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/41.jpg)
S.D.T. Applications
S.D.T. can be used in non-perceptualresearch, including memory experiments.
![Page 42: Signal Detection Theory](https://reader035.vdocuments.us/reader035/viewer/2022062217/56813af1550346895da36b5a/html5/thumbnails/42.jpg)
S.D.T. And Memory
The “new items”distribution
The “old items” distributionP
roba
bili
ty
Subjective Memory Strength (Arbitrary Units)
"No,I don'trecognizeit"
"Yes,I recognizeit"