TASK 1 - RT Flashcards

1
Q

cognitive psychology

A

= study of mental activity as an information-processing problem

  • identify internal processing (= acquisition, storage, usage) of information that underlies observable behaviour
  • adjusting/changing input and observing output –> what secrets are happening
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

cognitive psychology

- assumptions

A

1) mental representations: every person has individual, internal mental presentations of an object (e.g. when somebody tells you ‘ball’ we think of different things: object, description, mathematical formula)
- context dictates which representation is most useful
2) internal transformations: we do not directly perceive and act in the world: our perceptions, thoughts and actions depend on internal transformations/computations
- information is obtained by sense organs –> something secret happens (ability to comprehend and recognise it as something we already know/remember) –> choose appropriate response

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

mental representations

- Posner’s letter matching task

A

1) two letters are presented simultaneously in each trial
2) evaluate whether both letters are vowels, consonants or one and one
3) presses one button if letters are from the same category, and the other button if they are from different categories
= we have multiple representations of stimuli; response latencies reflect degree of processing required
–> 1. physical representations 2. phonetic representation 3. category representation

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

internal transformations

A

= recognition/comparison process operates either simultaneously or successively

a) parallel: reaction time should be independent of the number of items in memory set
b) serial: reaction time should slow down as memory set becomes larger –> time is required to compare an item with a large memory list than with a small memory list (4 items take longer to compare than 1)

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

internal transformations

- parallel processing

A
  • word superiority effect = participants are most accurate in identifying target letters when they are part of a word (word condition = fastest response time)
    1) stimulus is shown briefly (words, nonsense, X letter strings)
    2) participants are asked which of two target letters was presented (A or E)
    = when reading a list of words, we activate representations of words and their letters in parallel –> facilitates performance as both provide info about target
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

internal transformations

- serial processing

A
  • Sternberg’s task = compare sensory information with representations that are active in memory
    1) participant is presented with set of letters to memorise (1, 2, 4 letters)
    2) single letter is presented
    3) decide if this letter was part of the memorised set
    4) presses one button to indicate ‘yes’, it was part of the memory set or ‘no’
    = reaction time increased in a linear manner with set size –> support serial processing
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

constraints on information processing

A
  • participants are not able to compare target item to all items in memory set simultaneously (parallel)
  • specific constraint to system that you are investigating (task-specific constraints) or general processing constraint (people can only do a certain amount of internal processing at one time)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

constraints on information processing

- Stroop task

A

1) present list of words
2) ask to name the colour of each word as fast as possible (matching colour and word or not)
= faster response when words match the colour
1. representation corresponding to colour of each stimulus 2. representation of colour concept associated with each word –> activated, irrelevant to the task
- activation of representation based on word rather than the colour of the word appears to be automatic

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

reaction time

A

= speed with which a participant make their judgements

  • mental chronometry = the study of organisation + timing of mental processes
  • -> mental events occur rapidly and efficiently = we do not only measure correctness but also time = finer analysis of internal processes
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

reaction time

- operational definition

A

= time between onset of stimulus and overt response to that stimulus (e.g. button press)

  • trials are repeated to get a reliable mean RT (mRT)
  • participant intends to be as fast as possible without making errors
  • practically, allow errors and take mean time
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

reaction time

- theoretical definition

A

= minimum amount of time needed by the participant to produce a correct response

  • instructed and trained to raise validity: rather accept small proportion of errors than perfect performance to avoid risk that participant takes more time than actually needed (false results)
  • does not take into account errors, perfect assumption
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

reaction time

- PROS

A
  • time can be measured; actual variable of interest rather than an abstraction
  • highly sensitive to subtle differences between conditions
  • often displays an orderliness that encourages construction of models of the underlying information processing
  • ratio level: can be directly interpreted in a physical sense; meaningful 0 point, linear
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

reaction time

- CONS

A
  • only the end product of cognitive processing: the actual cognitive functions are hidden
  • -> infer backwards: assumptions are being made about contents of the black box/underlying
  • can’t use it for more complex tasks: they usually allow for more freedom leading to individual differences and variations within participants over time
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Donders’ subtraction method

A

= duration/change of stages

  1. construct two tasks that differ only in a single component of processing
  2. measure RT in both tasks
  3. subtract the RTs
  4. outcome is duration of the single component (influence of the independent variable
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

stages of mental comparison

A

1) encode/identify the visible target
2) compare mental representation of target + representation of items in memory
3) decide whether target matches one of memorised items
4) respond according to decision

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

Donders’ tasks

- task A

A

= simple RT task = one possible stimulus, one possible response
- e.g. every time you see the apple on the screen, press X; you have to start running when the signal comes
= sensory time + motor time

17
Q

Donders’ tasks

- task B

A

= 2-choice RT task = two stimuli demand different responses
- e.g. when apple appears on screen, press X; when melon appears, press Y
= sensory + discrimination + response selection + motor time

18
Q

Donders’ tasks

- task C

A

= go-gogo task = two possible stimuli, one possible response
- e.g. when apple appears on screen, press X; when melon appears, DO NOT press anything
= sensory + discrimination + motor time

19
Q

Donders’ tasks

- subtraction

A

RT(B) - RT(C) = response selection time

RT(C) - RT(A) = discrimination time

20
Q

Donders’ subtraction method

- assumptions

A

a) seriality: processing stages are carried out in a strictly serial manner; total RT is the sum of durations of individual stages
b) pure insertion: duration of all other processing stages stays the same when another stage is added/removed
c) knowledge of involved processes, by knowing tasks

21
Q

Donders’ subtraction method

- Sternbergs’ memory search paradigm

A

= proving serial processing: linear relation between target sample size and RT

  • -> set size function
    1) identify the visible target (L)
    2) compare L + with memory set (AKLM)
    3) decide whether L present in memory set
    4) respond: press yes
22
Q

set size function

A
  • slope: time it takes to scan/identify one item –> indicator of speed of memory search
  • intercept: time taken by anything else than memory (periphery, motor processing)
    a) self-terminating search: stop search when target is found (longer RT for neg. items)
    b) exhaustive search: scan whole memory set, decide whether hound or not (parallel lines)
23
Q

Sternberg’s additive factors method

A

= amount of stages

  • info is being processed in a series of successive functional stages (= mental modules): each stage performs some transformation on its input, produces some output that is passed on to the next stage
  • every manipulation has an effect on the different stages or on one in common –> addition or interaction
24
Q

Sternberg’s additive factors method

- additive factors

A

= factors are influencing separate stages

  • duration of one/more processing stages is increased but the output of that stage is not changed in quality
  • parallel slopes
25
Q

Sternberg’s additive factors method

- interactive factors

A

= factors influence the same stage, modify each other’s effect on a common stage

26
Q

Sternberg’s additive factors method

- assumptions

A

a) seriality
b) independence: effect of one factor will not depend on the other; if one stage takes longer, the next stage starts later but does not take longer

27
Q

Sternberg’s additive factors method

- challenges

A

1) accepting the null hypothesis: finding additivity boils down to acceptance of the null hypothesis concerning the interaction (however, “additivity” is the theoretically more important result: the discovery of a processing stage)
2) stage robustness: when the pattern of additivity/interaction does not change when a new factor is added (otherwise 3-way-interaction)
3) stages not sequential: AFM cannot be applied when two factors work on different features of a stimulus (e.g. location/identity) because then parallel processing and temporal overlap of stages
- RT would be shorter than sum of stage durations, violating AFM assumptions
- an interaction would be observed and AFM would overlook the existence of 2 separate stages

28
Q

continuous processing

A

= no separate stages

29
Q

problems

- practice

A
  • after practice: shorter RTs + less errors
  • only the intercept changes = periphery, motor process and NOT memory comparison (additive factors)
  • only get faster over time in unusual activities
30
Q

problems

- speed-accuracy tradeoff

A

= subjects cannot respond fast and make few/no errors

  • -> the faster/higher RT = more errors
  • instructions should encourage to act fast and to accept a few errors
31
Q

diffusion models

A

= model of decision-making

  • information accumulates continuously during the time between stimulus onset and response
  • -> accumulated information is represented by an internal counter which is driven in opposite directions by tiny bits of information supporting the different outcomes
  • information is driven up and down by random noise –> why response times and outcomes vary from trial to trial
  • explains distribution of Rts of correct responses and errors
32
Q

random walk

A

= randomly gather information pro and contra (randomly walk to top or bottom of graph)

33
Q

problems

- outliers

A
  • outliers have extreme effect on distribution of the values- avoid outliers: stress importance of consistency, (immediate) feedback to performance, brief experiments
  • take out outliers in first level of analysis
  • always right-skewed
34
Q

outliers

- level 1

A

= trial level

  • longer RTs: moments of inattentiveness
  • short RTs: premature answers without analysis of stimulus
35
Q

avoid level 1 outliers

A
  • reduce effects of potential outlier while eliminating as little as possible of the data of interest
    a) medians: insensitive to outliers; most central tendency = deleting upper and lower 50% of distribution
  • leads to biases if the number of observations is small and different across conditions
  • problematic in AFM
    b) trimmed means: delete small proportion (10% or fastest and slowest responses) –> take the mean of the remaining ones
  • choose percentage with care, avoid throwing out too many or too few
    c) C standard deviations: compute mean and standard deviation for each participant and condition –> delete RTs that deviate more than some number ‘C’ of standard deviations from the mean (C = 2.5 often used)
  • assumption of normality (typically not met for RT)
    d) fixed criterion: delete all RTs exceeding some (fixed) value –> compute the mean of the rest
  • lead to bias and loss of statistical power
    e) doing nothing: previous methods come with degree of arbitrariness
36
Q

outliers

- level 2

A

= subject level

  • one subject deviates from the rest (too slow RTs)
  • throwing out could be fraud: deleting unwanted results can lead to wanted results, although not representative of truth