Task 2 Flashcards

1
Q

What is a psychometric function?

A

-> It is a Graph , which is showing when the participant is capable of percieving something (Y) axis, while increasing physical intensity (tone X).

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

What are two response baises / reponse criterion ?

A
  • liberal responder

- Conservative responder

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

Waht is meant by liberal responder ? (Also explain how they would respond in a signal detction task)

A
  • response to even the slightest possibility of hearing a tone
  • In SDT = high hit high false alarms but low CR and low misses
  • Xc = left shifted
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is meant by conservitive responder ? (Also explain how they would respond in a signal detction task)

A
  • wants to be totally sure that she hears the tone before saying “yes.”
  • In SDT = low hit low false alarms But high CR and high misses
  • Xc right shifted
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

How to overcome the response criterion ?

A
  • Via the signal detection theory = introducing trails with no tone
  • Also introduce Payoffs = No error = 100 Euro
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

how does the signal detection tehory work ? (Method)

A
  • Two stimuli are present
    o Tone = signal
    o noise is all the other stimuli in the environment (often mistakes as signals)
  • The tone is in 50% of the trials present and 50% not present while the noise is always present.
    -> Randome order between tone and noise
    -> Meassures sensitivty of the subject!!
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

When is the tone easy detctable in a SDT task? Why is that bad ?

A
  • High intensity tone
  • Low level of noise (less pread) -> (the graphs become more narrow)
  • > In these trails we could not measure sensitivity we must always work around the threshold
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What types of noises do exist ?

A
  • > Internal noises

- > External noises

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is meant by internal noise ?

A

-> brain / neural activity

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

What is meant by external noise ?

A

-> random variations in the environment ( a fly / headphone moving)

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

How can an Error occur in a trial where the tone was present and in a trail where no tone was present ?

A
  • > Tone present = weak interanl respond ->Stmuli is weak and noise is weak
  • > Tone not present = Strong internal respond -> high noise
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What are the Genral response /decison criterion ?

A

-> XC = stands for answering convincively that a reponse is cuased by a signal and not by noise
Xc = internal response
Everything to the right of the XC = Yes caused by signal
Everything to the left of the Xc = “NO” caused by noise

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

What are the two responses which we can get in a SDT task with NO SIGNAL BEING PRESNET?

A
  • Correct rejection = no tone present -> answer no -> correct
  • False Alarm = no tone present -> answer yes -> wrong
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What are the two responses which we can get in a SDT task with A SIGNAL BEING PRESNET?

A
  • Hit = Tone present -> answer yes -> correct

- Miss = Tone present -> answer NO -> wrong

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

What is the overall goal of SDT ?

A
  • To identify the true sensitivy d’
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What does d’ stand for ?

A
  • Sensitvity
17
Q

What is an assumption of the STD ?

A
  • > We assume a normal distribution to make use of Z scores
  • > Internal noise has a specific distribution across trials (usually standard Gaussian = normal)
  • > Same noise distribution for N trials and S+N trials
  • > Signal and noise add up linearly (i.e., do not interact)
  • > Decision is based on two consecutive processing stages (sensory stage and decision stages)
18
Q

How do we compute d’ ?

A
  • P(FA) /Total number of no signal
  • P (H) / total number of signals
  • d’= Z (P(H)) – Z (P(FA))
19
Q

What are other options to identitfy if good sensitivty is present ?

A
  1. The overlap between the two graphs (best method)
    - > If the overlap is small = d’ large -> high sensitivity
  2. Or the distance between the two mean (peaks)
20
Q

How do we interpretet different d’ values ?

A
  • d’= 0 -> poor sensitivity / detectability -> just guessing
  • d’ > 0 -> good sensitivity / detectability -> the best
  • d’ < 0 -> good sensitivity / detectability -> but systematic error
21
Q

What does Beta (ß) stand for ?

A

-> strategy

22
Q

How do we compute the strategy / beta?

A
  • > Z(P(FA)) and Z(P(H)) as before (see previous slides)
  • > Y(Z(P(FA))) = height of N distribution at Xc
  • > Y(Z(P(H))) = height of S+N distribution at Xc
  • > β = Y(Z(P(H))) / Y(Z(P(FA)))
23
Q

What is a important computing rule ?

A
  • > If we know What P (CR) is then we know P (FA)

- > 1 - P (CR) = P (FA)

24
Q

How do we more simply identify Beta ?

A
  • > It is the relative height at the two distributions (graphs) at the criterion (Xc)
25
Q

How do we interpretet the Beta ?

A
  • Beta is large = more conservative

- beta is small = more liberal

26
Q

What is an important rule of STD ?

A
  • > If you change your response criterion (Xc) the only thing which change is the type of wrong answers
  • > The amount of wrong answers stays the same
27
Q

What is meant by the optimal beta ?

A
  • > Beta optimal = the strategy with the lowest coast but the highest benefits
  • > alos taking into account the probability of signal occurring
28
Q

What is meant by sluggish beta ?

A
  • In real life, participants adjust their criterion according to the payoffs, but the criterion is not perfectly optimal
29
Q

Explain the pay off matrix:

A
  • > If I put high payoffs only on CR -> Xc moves to the right -> less yes
  • > If I highly punish only FA -> Xc moves to the right-> less yes
  • > If I highly punish only misses -> Xc moves to the left -> more yes
  • > If I put high payoff only on a Hit -> Xc moves to the left -> more yes
30
Q

What are roc cruves ? And what can they tell us ?

A
  • > The perceiver’s responses for a whole range of strategies
  • > the sensitvity
  • > Strategy used
  • > Plots P (H) on the y axis and P (FA) on the x axis)
31
Q

Based on the Roc curves how do we determine the sensitvity ?

A
  • > Early increased and bowed = good sensitvity
  • > Linear = guessing
  • > Late increase = good sensitivity but systematic errors
32
Q

Based on the Roc curves how do we determine the Strategy ?

A
  • > Low P (H) + Low P (FA) = conseravive
  • > medium Low P (H) + medium Low P (FA) = smaller conservative
  • > medium high P (H) + medium high P (FA) = smaller liberal
  • > High P (H) + High P (FA) = liberal
33
Q

How can we identify if lots of noise is present ?

A

-> the higher the peaks in the two graphs the more noise is present

34
Q

What type of components of the subject decison making are refelcted in a yes / no response

A
  • > Sensetvity = SDT

- > strategy = Payoffs

35
Q

What is in general the defintion of the SDT ?

A

-> Decision making under uncertainty (close to threshold)