02-03. Scaling & signal detection Flashcards

1
Q

Psychophysical scaling def?

A

describe relationships between subjective experience and stimulus intensity

Scale —> compare 2 things
—> subj. ex. & phys, intensity

3 laws!!

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the 3 laws called?

A
  1. Weber’s Law
  2. Fechner’s Law
  3. Steven’s Power Law
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Weber’s Law? (explanation)

A
  • JND gets bigger for increasing stimulus magnitudes
  • (bigger stim. size = bigger JND)
  • JND is CONSTANT proportion of intensity of standard
  • fails at low intensities and near the threshold
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Weber’s law equation?

A

JND = k∗I
Weber fraction: k = JND/I

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Fechner’s law assumptions

A
  1. Weber’s law is right
  2. Each JND feels the same (the JND is the basic unit of perception)
  • since Weber’s law fails, Fechner’s law does too
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Fechner’s law? (explanation)

A
  • EQUAL increments in phys. int. produce SMALLER inc. in perceived magnitude
    –> adding a candle to a dark room vs adding a candle to a room with 100 candles
  • PERCEIVED INTENSITY increases as a LOG function
  • the change in physical intensity needed to produce one JND increases logarithmically as well (??)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Fechner’s law equation?

A

S = k ln I/I₀

S=perceived intensity
k=weber fraction
ln = nat. log –> (IMPORTANT)
I=physical intensity
I₀=phys. int. @ threshold

graph is log (decreasing slope)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Steven’s power law? (explanation) what sort of measurement does it use?

A
  • says power function describes it better for more stimulus modalities
  • uses magnitude estimation
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Magnitude estimation?

A
  1. present standard stimulus with arbitrary number
  2. present other stimuli and subject assigns number relative to standard

ex. standard line is 100, following lines are described as 200, 50, 75, etc

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Steven’s power law equation?

A

S = c * I^n

S=perceived
I=phys
c=constant
n=exponent

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Possible exponent values?

A

n > 1… perceived int. increases at INCREASING rate
- ex electric shock

n = 1… perceived int. increases at CONSTANT rate
- ex. line length

n < 1… perceived int. increases at DECREASING rate
- ex. brightness, auditory stim.
- (this does fit Fechner’s law)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Signal detection theory? Experiment?

A

permits the measurement of a person’s sensitivity (to weak signal) while unaffected by their biases

Experiment:
- present weak stimulus with NOISE

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Signal detection experiment:

A
  1. use 1 fairly WEAK stimulus
  2. on every trial, present NOISE
  3. Some trials have a signal embedded in the noise
  4. Subject decides if they detect signal or not
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Possible responses for sig. detection

A

HIT (signal + yes)
FALSE ALARM (no sig + yes)
MISS (signal + no)
CORRECT REJECT (no sig + no)

You only need to know Hits and False Alarms to have all the information you need

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Receiver operating characteristic

A

graph where you can plot the proportion of hits and false alarms

shows all possible isometric curves

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Isometric curve (on ROC)

A
  • Indicates an individual’s sensitivity by the curve that their data point (FAR, HR) falls on
  • an individual can get 2 different data points ON THE SAME CURVE

top right corner is the most sensitive

diagonal line is random responses, completely insensitive

17
Q

Payoffs/bias affect isosensitivity curves?

A
  • if you tell someone there’s gonna be a signal most of the time, they’re heavily biased to say yes (high false alarm rate)
  • if there’s a big cost to missing (money, or in diagnosing diseases, etc), people are heavily biased to say yes as well
  • payoffs/biases MOVE where a person’s dot is on the SAME isosensitivity curve!!
18
Q

Sensitivity def
True ___ rate?

A

True positive rate (Vertical axis)

How sensitive they are to detecting the signal

19
Q

Specificity def

A

True negative rate (horizontal axis)

How accurate they are in saying no when there isn’t a signal