Task 1 Flashcards
Letter-matching task by Michael Posner - method
-2 letters presented simultaneously
-task: are they both vowels, both consonants or different?
-> press buttons
> Variables: capitalized/non capitalized; vowel/consonant
5 conditions:
1) physical-identity condition: 2 letters same
2) phonetic-identity: 2 letters same identity, but one is capitalized
3) 2 types of same-category: 2 letters are different members of same category; 1) both letters are vowels, 2) both letters consonants
Letter-matching task (Posner) - 3 representations derived:
>
Representation based on physical aspects of the stimulus (visually derived from shape presented on screen)
>
Representation corresponds to the letter’s phonetic identity (reflects fact that many stimuli can correspond to same letter, e.g. A and a)
>
Category to which a letter belongs (A and E => activate representation of ‘vowel’)
Conclusion letter matching task Posner
physical representations are activated first, phonetic representations next, and category representations last
Internal transformations
- Mental representations undergo transformations
- Taking action often requires that perceptual representations be translated into action representations in order to achieve a goal
- Information processing is NOT simply a sequential process from sensation to perception to memory to action
Sternberg’s task – characterizing transformational operations
Task:
- Comparing sensory information with representations that are active in memory
- First: participants are presented with set of letters (1/2/4) to memorize
- Then: single letter is presented -> decide if letter was part of the set
To respond in stern bergs recognition task, the participants must engage in 4 primary mental operations:
1) Encode: identify the visible target (sensory processing)
2) Compare mental representation of target with representations of items in memory (stimulus discrimination)
3) Decide (response selection)
4) Respond (motor response)
Sternberg’s task recognition process might work in 2 ways…:
1) Parallel process => if comparison process can be simultaneously for all items, THEN RT should be independent of the number of items in the memory set
2) Serial process => RT should slow down as the memory set becomes larger
> Supported by Sternberg’s results
Propositions additive factors method Sternberg
1) Information is processed in a series of successive functional stages -> each stage performs transformation on its input and output is passed to next stage
2) If we use manipulations to increase RT -> duration of one (or more) processing stages is increased, but output of that stage is not changed in quality
> Output of a stage is constant across conditions and does not depend on its own duration
> Later processing stages cannot compensate for delays earlier on
> Effects of 2 factors should be additive, should not interact in a statistical sense
3) if 2 manipulations mutually modify each other’s effect, if they interact -> must affect some stage in common
Additive Factors Method applied: Memory Scanning and Stimulus Degradation:
- show digits
- afterwards: present digit and indicate which was the digit from before
factors: set size and stimulus degradation:
1) intact: stimuli clearly perceivable digits
2) degraded: checkerboard pattern superimposed -> increased RT
Theories additive factors method applied - memory scanning and stimulus degradation
Theory I:
- 2 independent stages: a) identification, b) serial comparison to representation of each item of the memorized set
- Degradation increases only duration of stage a independent of set size
- Intercept (duration: how long it takes to compare 0 objects, this takes longer for degraded than not degraded) of set size function is increased, but not the slope (additivity -> different stages are influenced independently)
- Data supports this theory
Theory II:
- Only a single stage compares degraded code to representation of each memorized item in a set -> takes longer than intact stimuli
- No effect on the intercept, but slope
- Degraded stimulus is compared to memory items, increasing stage b for every one of those items (interaction -> stage in common)
Additive Factors Method: Challenges and limitations:
1) additivity -> normally we don’t like to accept a H0
2) Stage robustness: (= pattern of additivity/interaction among factors does not change if these factors are combined with a new factor) -> we can’t be sure that this is the case
3) Stages not sequential: in some circumstances, the nature of factors is such that they plausible affect 2 stages that are logically not sequential
> Example: 2 factors work on different features of the stimulus (location and identity)
> If they can be processed in parallel, then stages would show temporal overlap -> RT would be shorter than the sum of stage durations -> violating AFM assumptions
Word superiority effect – Reichner
-Although memory comparison appears to involve a serial process, much of the activity in our mind operates in parallel (as opposed to serial like sternbergs model)
-Stimulus is shown briefly -> participants is asked which of 2 target letters was presented (e.g. A/E)
-Participants are most accurate when stimuli are words (= word superiority effect)
-Suggest that we do not need to identify all letters of a word -> representations of the individual letters and
entire word are activated in parallel for each item
Assumptions of the Subtraction method: (cons of the method)
1) Seriality: processing stages are carried out in a strictly serial manner (but there is evidence that processing can work in parallel)
2) Pure insertion: duration of all other processing stages remains same when stage is added/removed -> stages are independent of each other -> it may be possible to construct tasks in which this assumption is valid, in many cases (Donders original tasks) it is likely not
3) The method presupposes detailed knowledge about the processes involved in a task before we can construct a good comparison task
Set-size function , Sternberg’s task
Mean RT = 397.2 + 37.9 * s
> Extra time needed to scan a single additional item in memory
> Linear = time to scan each item does not depend on the total size of the set
> Example of ‘orderliness’ that RT data can display
> Intercept (397.2) = theoretical time needed for scanning zero items -> do not depend on set size
> Slope (37.9 = time to compare one item in your head to memory) = very similar for positive (yes) and negative (no)
> Apparently, participants continue searching the set, even after finding the match -> exhaustive search in both cases
> Possible explanation: serial processing or limited capacity parallel processing
self-terminating search
positive response: as soon as match has been found, search ends
- Reaction time should be smaller for positive items compared to negative items
- IF it would be like that, the slope for positive responses should amount half of the slope for negative responses (bc. On average, a match should occur after searching half of the items, in Sternberg’s memory scanning paradigm)
