Problem 1 Flashcards
Cognitive neuroscience
Study of mental activity as an information-processing problem
–> seeks to identify internal processing of info that underlies observable behavior
Letter matching task
Posner
- 2 letters are presented simultaneously in each trial
- Evaluate whether they are both consonants, both vowels or a mix
(3. Adding of 2 other conditions where 2 letters are the same (physical) or same letter but one capital other lowercase (phonetic))
Result of posners experiment
We derive multiple representations of stimuli that are based on
- the physical aspects
- -> AA - on the letters identity
- -> Aa - categorization
- -> vowel or consonant
THUS: different response latencies reflect the degrees of processing required to perform this taks
Dependent variable (letter-matching task)
Reaction time
–> the speed with which the participants make their judgments
Participants must engage in 4 primary operations for sternbergs task.
Name them
- Encoding
- Comparing
- Deciding
- Responding
Donders subtraction method
- Construct 2 tasks that differ only in a single component of processing
- Measure RT in both tasks
- Substract the RTs
- Outcome is the duration of that single component
e. g. like letter matching task
Reaction time (Operational definition)
Refers to the time interval between the onset of a stimulus and the overt response to that stimulus
–> here button press
BUT: assumes that the participant intends to be as fast as possible w/o making errors
2-choice RT
B type Donders task
2 possible stimuli and each of them demand a different response
–> reaction times are longer here than A
Go-nogo Task
C type Donders task
2 possible stimuli and one possible response
–> repsonding to go but not logo
Assumptions of Donder’s method
- Seriality
- -> processing stages are carried out in a strictly serial manner - Pure insertion
- -> when an extra stage is added/removed the duration of all other processing stages remain the same - Presupposes detailed knowledge about the processes invoved
Why was the assumption of pure insertion criticized ?
Appears to be open to empirical verification
–> although it may be possible to construct tasks in which this is valid, in many, like the donders original one it is likely not
Sternbergs memory search task
Serial processing
- Set of 1-6 letters is presented, to memorize
- One letter is then demonstrated after the rest disappeared
- Participant has to decide whether that letter was in the initial set
- Indicate yes or no
–> Comparing sensory info (stimulus) with representations that are active in memory
e.g.: How fast can we search memory?
Factor
Independent variable in an experiment
e.g. gender
Levels
Set of values of a factor
e.g. gender = male + female
Main effect
Is the Mean of all the simple effects of a factor
Simple effect
Is an effect of one factor at one particular level of a factor
Interaction
If 2 manipulations modify each others affect, they will also affect a stage in common
Additive effects
If 2 manipulations affect different stages, then the effect of one will not depend on the other
Main assumptions of AFM
- Seriality
- -> thus total RT is simply the sum of the stage durations - If we use some experimental manipulation to increase RT, then the durations of one or more processing stages is increased
- -> BUT output of the stage is not changed in quality (constant + independency of stages) - Later stages of processing cannot compensate for any delays earlier on
- -> increased duration of some stage is simply propagated through all following stages
BUT: Does NOT assume pure insertion
Stage robustness
Obtained if the pattern of additivity or interaction among factors does not change if the factors are combined with a new factor
Speed accuracy tradeoff
There is a tradeoff in how fast a task can be performed vs how many mistakes the participant makes
–> because the faster a participant responds, the more errors they make
Diffusion model
Provides a theory that explains the distribution of RTs of both correct and error responses in a 2-choice RT task
- ->
1. as info accumulates between stimulus onset and the response
- represented by an internal counter, reacting to the received information supporting different outcomes (correct vs. incorrect) by shifting in opposite directions
- when the counter hits the lower or higher threshold a certain response (a or b) is elicited
- random noise alters the counter
- -> explains the variation of RTs in each trial, while the stimulus remains the same
Word superiority effect
One is most accurate in identifying target letters when they are part of a word
–> we do not need to identify all the letters of a word before recognizing the word
Reaction time (Theoretical definition)
Refers to the minimum amount of time needed by the participant to produce a correct response
Advantages of RT
- Mental processes have no physical properties but time can be measured
- Highly sensitive to subtle differences btw conditions
- It is the actual variabel of interest
Disadvantages of RT
- Reflects only the end product of cognitive processing whereas the actual functions are hidden
- Cant use it for more complex tasks
Sternbergs additive factors method
AFM
Discovers independent stages, thus how many there are and what they do
–> assumes that in a sequence of stages (=mental modules) eachh does some operations and hands output to the next
How may one correct the speed accuracy trade-off ?
Since there is no evidence that the participants actually use the minim possible amount of time to react:
–> Running the same task multiple times and give a different speed accrue instruction every time
How accurately can one identify outliers in RT measurements ?
This is difficult since one cannot discriminate between “true” slow RTs and outliers
–> should thus stress the importance of
- consistency
- practice
- brief tasks
- feedback
to keep people alert
Assumptions of the diffusion models
- Info accumulates continuously during the time between stimulus onset + response
- Info is represented by an internal counter which is driven in opposite directions by tiny bits of info supporting the different outcomes (correct vs incorrect)
- There is random noise in this info that drives the counter
–> explains the variation in the outcomes even when the stimulus is the same
Random walk
Simulating the random noises the drive the internal counter with a computer program
–> to obtain simulated RT distributions, that mimic actual ones as closely as possible
How do you investigate SAT?
- Sample the whole SAT function in each condition, by repeating the experiment many times under different speed-emphasis instructions (takes a lot of time and effort).
- Fit both RT and accuracy by a single mathematical model
- -> If the model’s assumptions are fulfilled, the best fitting parameter values of the model accurately describe the characteristics of the underlying processing
THUS: diffusion models: IMPORTANT
